Gold Fish 3

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

/ Gold Fish 3 / goldfish_volume_3.bin / files / dev / asm / barflydemo / docs / barfly.man < prev next >

Wrap

Text File | 1995-09-05 | 181.8 KB | 6,421 lines

This file documents Barfly, a development system for AmigaOS V2.04 or newer. Barfly 1.0 An Intuition controled Debugger and Optimizing Assembler Copyright (c) 1989-94 Ralph Schmidt - Shareware - Basic Informations ****************** Copyright and other legal stuff =============================== Copyright (c) 1989-94 Ralph Schmidt Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies. No guarantee of any kind is given that the programs described in this document are 100% reliable. You are using this material at your own risk. The author *can not* be made responsible for any damage which is caused by using these programs. Permission is granted to include this package in Public-Domain collections, especially in Fred Fishs Amiga Disk Library (including CD-ROM versions of it). The distribution file may be uploaded to Bulletin Board Systems or FTP servers. If you want to distribute this program you *must* use the original distribution archive `Barfly.lha'. *None* of the programs may be included or used in commercial programs unless by written permission from the author. *None* of the programs may be modified in any way. This especially includes changing the copyright notices or removing any of the Shareware restrictions. *None* of the programs may be used on any machine which is used for the research, development, construction, testing or production of weapons or other military applications. This also includes any machine which is used for training persons for *any* of the above mentioned purposes. *None* of the programs may be used by more than the registrated owner. Registration ============ As you may have noticed in the copyright I'm working for five years at Barfly. It has always consumed and will continue to consume a large amount of my time. I cannot afford just working for fun. Thus, I decided to release Barfly as Shareware. I had already tried to release Barfly as a commercial product but the story behind it is more than sad. To sum it...german Amiga software companys aren't worth any time...`they suck'. Some people may think the price is too high for a Shareware product but i think that BAsm is as powerful as the 2 main available commercial Assemblers...if not more powerful if you compare the speed and the optimize functions;there's no commercial Debugger available that can compete with BDebug. I've used Barfly myself for commercial Amiga applications. Cyberstorm 060, Z3-Fastlane device, CDRive, SCSIConfig,... The unregistered version of Barfly pops up the About requester at the start and has some functions disabled: Assembler: o only 8192Bytes large code possible o the Section commands aren't available Debugger: o Only 1 Window per Object o Enforcer,CyberGuard Catch not available o Task Catch not available o Crashed Task Catch not available o Limited Step count(about 150-200 Steps) Registered users will be shipped a disk with the newest public release of Barfly, along with a personalized, so-called "keyfile". It enables all the missing features and disables the Shareware reminders. This keyfile will work with all future releases of Barfly, so you can simply download the latest version from your local bulletin board without having to wait weeks for your update passing through the slow mail channels. The keyfile `must not' be distributed in any way. The fee for a Barfly registration is 70.- DM (D-Mark), 70.- SFr (Schweizer Franken), 230.- FF (French Francs), 50.- US$ (US Dollar) The fastest, cheapest and easiest way to register is put the money together with the filled registration form into a letter and send it to Please allow 2-8 weeks delivery for the registrated version. Currently I am very busy with my civil service and working for Phase5 so please be patient. Ralph Schmidt Kleiner Hellweg 4 33154 Salzkotten FR Germany Phone: +49-5258-5637 E-Mail: laire@mises.uni-paderborn.de Irc: Laire on #amiga,#amigager Introduction ************ Purpose ======= BDebug is an Intuition controled multi-task system Debugger for OS 2.04 and newer. You can use BDebug to debugging your programs, catching tasks, reseach Enforcer and CyberGuard hits, follow Source-Level Informations and other advanced functions. The Debugger supports assemblers, SAS C, Dice and GCC. Some of BDebug's features are: * font-sensitive, resizable and Style Guide compliant GadTools GUI * Object-Oriented that results in a low learning curve * Supports 68000...68060 and the FPUs * Can debug multiple tasks at the same time * Not limited by the amount of window objects. * highly configurable * keyboard support BAsm is a very fast optimizing Assembler for OS 2.04 and newer. Some of BAsm's features are: * 68000-68060, 6888x * Very Fast * Include and Incbin Cache * Strong Optimizer with Multi-Pass Optimizing * High Level Macros * ARexx * Supports OS 2.04 and OS 3.0 Hunks * SAS D1 Source Level Format System requirements =================== Barfly requires Amiga operating system version 2.04 or better. Kickstart 1.3 is *not* supported; this operating system is considered obsolete. Barfly requires at least one megabyte of RAM to run. A hardisk or a faster CPU is not required but increase performances and comfort, of course. Installation ============ It is really easy to install Barfly: 1. Copy the the binary and the icon (called "BDebug" and "BDebug.info") to any directory. 2. Copy the the binary and the icon (called "BAsm" and "BAsm.info") to any directory. Then copy the supplied configuration files from "s/Barfly/#?" to "S:" directory of your system partition or create env:Barflypath with the path to a directory that contains Barfly/#?. Other topics ************ Updates ======= Whenever a new release of Barfly gets released, I will post some information in the appropriate newsgroups of some electronic networks. The new archive will soon be available on many bulletin boards and on all `AmiNet' FTP servers. Major releases will also come with some PD disks, especially on Fred Fish's collection. As mentioned above, registered users will neither need a new keyfile nor a special personalized program version. They can use all new features immediately. Support ======= If you have some questions, comments, suggestions or even flames, please feel free to contact me at one of the following addresses. If you send your letter via e-mail, there's a good chance for getting a quick reply. Snailmail: Ralph Schmidt Kleiner Hellweg 4 33154 Salzkotten FR Germany Phone: +49-5258-5637 E-Mail: laire@mises.uni-paderborn.de Irc: Laire on #amiga, #amigager History ======= * 0.0 - 1.0 not released Future ====== Here are some ideas for future versions of BDebug: * Full Source Level support for GCC and perhaps SAS. * Better BDebug Arexx support...the current one is a bad excuse. * Mungwall Trace methods. * Automatic Refresh of some Windows(Task Window...) * Amigaguide file mode in the autodocs functions. * Highlight changed registers * Better documentation. * BAsmOption for the easy BAsm options configuration. * Other things i'm too lazy to mention now. Important: There is absolutely NO guarantee that these features will ever be implemented. So don't be disappointed, if they aren't in the next version. Acknowledegments ================ Thanks must go to: - Dirk Leschner, Frank Jakobs for being my best friends - Matthias Scheler for his manual design and Filer. - S.Schaem, Børge Nøst, Alexis Wilke, Michael B. Smith, Marc Heuler for their superb betatesting efforts - Mike Schwartz for a lot suggestions to improve basm. Sad he has left the Amiga community 2 years ago. - Stefan Becker for Toolmanager and being a nice guy. - Christoph Wolf for DynamiCache and being a nice guy. - Brian Cerveny(Redwine) for Grapevine and being a nice guy. - All my IRC friends. for many great hours. Thanks! Andrew Denton(Guardian), Kenneth Dyke(Nyx), Bill Coldwell(Cryo), Brian Cerveny(Redwine), Joseph Hillenburg(xterm), Scott Ellis(ScottE), Chris Wichura(Caw), John Wiederhirn(John\_W), Mike Schwartz(mykes), Markus Illenseer(ill), Petra Zeidler(stargazer), Michael van Elst(mlelstv), Holger Lubitz(holgi), Ralph Babel(rbabel), Seth Harman(Budha) and a lot guys i haven't mentioned here. - Chris Schneider and Urban D. Mueller for some suggestions 2-3 years ago and installing Aminet. - Michael "billy" Böhnisch for his cleanup on MakeBarflyFD 2-3 years ago. - Steve Wright for designing the icons 2 years ago. And of course to all the other Beta testers and registered users. The V40 Includes can be ftp'ed from the FTP-Server. ftp.rz.uni-wuerzburg.de: pub/amiga/frozenfish/bbs/com A superb Linker "lk" by Alex Wilke should be soon availble on Aminet. BDebug 1.0 An Intuition controled Debugger Copyright (c) 1989-94 Ralph Schmidt - Shareware - Usage of BDebug *************** The Command Window ================== Debugger Philosophy =================== BDebug is a multitasking Debugger that supports the Motorola processors 68000...68060 and 68881...68882. The Debugger allows to debug unlimited tasks parallel. Because of the Debugger's complexity BDebug was designed in an object-oriented way to allow an easy and comfortable way to use it. The register window `REGWindow' is the Root class of the task object that can be expanded by several subclass windows. Every subclass window has privat menus and inherits the public menus of its father object. Debug Methods ============= The Debugger offers a variety of different Debug methods that can be activated by menu or gadget. Debug Task .......... is used to select a task you wann to debug. If you doubleclick on a task a `REGWindow' and a couple of information windows opens. Which type and how many are opened depends on the current configuration. After the task could be stopped the contents of the `REGWindow' and all other information windows gets refreshed. If the task is in the `Wait' state the task is stopped when it gets a signal. Task Listview Layout Taskaddress & Priority & Status & Name Process Listview Layout Taskaddress & Priority & Status & Name[CLI Number,CLI Name] You should know what task you can stop and what kind of task should never be stopped. For example the Input.device should never be stopped. Debug File .......... is used to load and stop a program. This function is equal to the bdebug cli startup with the exception that you can enter the parameter in a requester. If no error occur all configurated windows are opened and the PC stops at the defined programstart breakpoint that normally points to the first command of the program. Debug Next Task ............... is used to debug the next task that is opened. The Debugger waits until another task is created by AddTask and a couple of information windows opens. Which type and how many are opened depends on the current configuration. After the new task was caught the pc points to the beginning of the task and `Catch Next Task' is disactivated. To catch a program that is started from the WV you have to use `Debug Next Task' to catch the WB Startup Task `WBL' that starts the program. Now you have to activate `Debug Next Task' again and let the current task run. After a short time the task `WBL' ends and the program's task is caught. You should avoid to start a new task between the 2 `Debug Next Task' phases because it's easy to catch the wrong one. You should notice that `AddTask' is patched and points to a new routine. Thus you should be careful with programs that also patch `AddTask'. Furthermore it's useful to know in what sequences these patches have to be removed. The Debugger can only be closed when all patched system function that were installed after the Debugger was started are removed. If you start a program in the shell without `c:Run' no new task is created. Instead the program is run as a subroutine in shell's task so you can't catch the task that easy. Debug Crashed Task .................. is used to catch tasks that crash so you track down the bug location a lot easier. If the system itself doesn't run anymore you shouldn't expect that bdebug still runs because it depends on a working system. If a task crashes and the option `Debug Crashed Task' is activated a couple of information windows opens. Which type and how many are opened depends on the current configuration. `Task Held Requester'. Catch Enforcer Hit .................. is used to tell the Debugger to stop the task it controls when an Enforcer or CyberGuard hit happens. Unfortunately the Debugger can't always stop the task at exactly the same location where the hit happened. Mostly the hit command is 1-2 instructions above the stopped task's PC. This function needs Enforcer V37.x by M. Sinz or CyberGuard V1.x and it must be installed before BDebug is launched. Please read the documentation. Select Display Mode ................... This function allows you to choose Screen Mode for the Debugger. How to start ? ============== If you only want to debug a program you have to start bdebug with the program's name and parameters or by using `Debug Program' in the command window. Another method is to move a program's icon on the command window or specify BDebug as in the icon as the `DefaultTool'. `Name:' BDEBUG - The CLI Startup `Synopnis:' BDEBUG [?] [<Program> [Argument] ] `Function:' BDebug activates the Debugger, loads and stops the optionally entered program. If it can find a local config file with the suffix *.bdebug it loads it. `Inputs' * ? shows an information message * program name If no program name is entered BDebug looks for env:BDebugProgram and loads the program instead that the env: points to. * argument line of the program. If there are spaces in parameters you have to enclose the argument with `""'. Usage of BDebug *************** The Register Window =================== Register Window --------------- The register window is the most important control layer of the Debugger and every debugged task has one. You can link unlimited other information windows to the `REGWindow' or you can tell the Debugger to give up controling the task. In the title line of the window you can see the ID number of the task, so you can recognize what information window belongs to this task. Furthermore the title line also contains the task address, the state, `RUN', `WAIT' or `STOP' and the name of the task. The task is in the `RUN' state if the task has the control at the moment instead of the traphandler; the task is in the `STOP' state when the task waits and the traphandler controls what happens. The task is the `WAIT' state only when the Debugger has to wait to catch a task by a `Debug Task'. In the upper area of the window you see the normal data and address registers where the address register also have additional information fields. To change a register or to watch memory where the register points to you only need to doubleclick on the register or on the register's memory contents. Furthermore you may change other usermode registers by this method. Supervisor register can't be changed because that doesn't make much sense for a system Debugger. `68000 Registers' D0=xxxxxxxx yyyy A0=xxxxxxxx Arg1 Arg2 [!] D1=xxxxxxxx yyyy A1=xxxxxxxx Arg1 Arg2 [!] D2=xxxxxxxx yyyy A2=xxxxxxxx Arg1 Arg2 [!] D3=xxxxxxxx yyyy A3=xxxxxxxx Arg1 Arg2 [!] D4=xxxxxxxx yyyy A4=xxxxxxxx Arg1 Arg2 [!] D5=xxxxxxxx yyyy A5=xxxxxxxx Arg1 Arg2 [!] D6=xxxxxxxx yyyy A6=xxxxxxxx Arg1 Arg2 [!] D7=xxxxxxxx yyyy A7=xxxxxxxx Arg1 Arg2 [!] USP=xxxxxxxx SSP=xxxxxxxx PC=xxxxxxxx SR=xxxx * [!] shows if the address register points on an odd address. * if the address register points on an illegal memory area the char `*' is shown in Arg1 and Arg2 to avoid crashes by reading non readable io-addresses. You can config the readable memory areas. * if the address register points on the following system structures the name of the Node is shown in Arg1 and the Name of the structure in Arg2. * Library * Device * Port * Task * Resource * MemHead * if the address register points to the custom chip area the name of the register is shown in Arg1 and the ID-Word CUSTOM is shown in Arg2. Custom register map is $dff000-$dff200. * if the address register points to a symbol the symbol and the contents is shown. * Otherwise 8Bytes of the memory are shown where the register points to. The 8 Bytes are shown hexadecimal in Arg1 and ascii in Arg2. `68010 Registers' VBR=xxxxxxxx SFC=xxxxxxxx DFC=xxxxxxxx `68020 Registers' VBR=xxxxxxxx SFC=xxxxxxxx DFC=xxxxxxxx MSP=xxxxxxxx ISP=xxxxxxxx CACR=xxxxxxxx CAAR=xxxxxxxx `68030 Registers' VBR=xxxxxxxx SFC=xxxxxxxx DFC=xxxxxxxx MSP=xxxxxxxx ISP=xxxxxxxx CACR=xxxxxxxx CAAR=xxxxxxxx CRP=xxxxxxxxxxxxxxxx SRP=xxxxxxxxxxxxxxxx TT0=xxxxxxxx TT1=xxxxxxxx TC=xxxx PSR=xxxxxxxx `68040 Registers' VBR=xxxxxxxx SFC=xxxxxxxx DFC=xxxxxxxx MSP=xxxxxxxx ISP=xxxxxxxx CACR=xxxxxxxx URP=xxxxxxxx SRP=xxxxxxxx TC=xxxx PSR=xxxxxxxx ITT0=xxxxxxxx ITT1=xxxxxxxx DTT0=xxxxxxxx DTT1=xxxxxxxx `68060 Registers' VBR=xxxxxxxx SFC=xxxxxxxx DFC=xxxxxxxx CACR=xxxxxxxx PCR=xxxxxxxx BUSCR=xxxxxxxx URP=xxxxxxxx SRP=xxxxxxxx TC=xxxx PSR=xxxxxxxx ITT0=xxxxxxxx ITT1=xxxxxxxx DTT0=xxxxxxxx DTT1=xxxxxxxx Type Information xxxxxxxx [Symbol] Mnemonic operand1[,...]] (EA): [Address1=Contents]...[Address2=Contents] In the `(EA)' line you can see the addresses and their contents the current command accesses. The contents of illegal addresses aren't shown. Local Menus ----------- * Close Window closes the `REGWindow', all connected windows and disactivates the Debugger for this task. To disactivate the Debugger you have to choose if the task should keep running so it's the task's business to stop. Furthermore you can end the task by runing the cleanup routine of the task or just removing the task from the list but this can cause sideeffects you can't always oversee. If the task is a process then the Remove option is equal to the Cleanup option. If it's only a task the Cleanup option is equal to the Cleanup option. Note that the Remove option doesn't free any resources of the task. You should really know what you're doing if you, for example, remove a task from the system. * ZOOM Windows expands all windows of the task. * Log File activates or disactivates the loging of the register and PC changes. * Big View shrinks the `REGWindow' to a 68000 register layout or expands it back to the full layout. * Open DissWindow opens a DissWindow with the configured dimensions. * Open MemWindow opens a MemWindow with the configured dimensions. * Open FPUWindow opens a FPUWindow with the configured dimensions. The menu is only available if a FPU is installed. * Open BreakWindow opens a BreakpointWindow with the configured dimensions. * Open CoppWindow opens a CoppWindow with the configured dimensions. * Open StructWindow opens a StructWindow with the configured dimensions. * Open SnoopWindow opens a SnoopWindow with the configured dimensions. * Open WatchWindow opens a WatchpointWindow with the configured dimensions. * Open ChecksumWindow opens a ChecksumWindow with the configured dimensions. * Save Window Settings saves the positions and count of all window the current task controls. The saved file then contains the appropriate commands you have to enter yourself into the configuration file. Because of the Debugger's window concept it doesn't make sense to save a complete configuration file. Public Menus ------------ * Step 1 runs the current command and stops the task afterwords. * Step X runs X commands and stops the task afterwards. * Step Debug Line runs commands until the PC encounters another source line. If the PC is outside the program or if no debug informations are available a single step i used. The command enters subroutines. * Trace Debug Line is similar to `Step Debug Line' with the exception that it runs subroutines. * Trace over Calls runs the current command or subroutine and stops the task afterwards. Depending on the configuration and the memory area a breakpoint or single steps are used. If a crash happens in certain program parts you should remove the command `Tracebreak' from the configuration. You should avoid to use `Tracebreak' in Libraries and Devices which are located in the ram. If another task accesses the routine at the same time you can expect an illegal exception. Some Amiga MMU Setups don't like programs writing to the kickstart rom. For example when breakpoints are set. * Trace X over Calls runs X commands or subroutines and stops the task afterwards. Depending on the configuration and the memory area a breakpoint or single steps are used. * Trace Work is similar to the command `Trace over Calls' with the exception that all commands are run. This function is useful to trace loops. Example: moveq #10,d0 0$: dbra d0,0$ If you use the function on the command `dbra' the Debugger sets a breakpoint after the dbra and runs the task. It drops back to single step when it hits a `Jmp', `bra', `rts'.... * Trace over OS-Calls runs the current command or the OS function and stops the task afterwards. * Trace on Flow stops the task when a PC direction occurs. This means the PC is stopped when it hits a `Jsr', `Jmp', `bcc', `rts'.... * Trace on Address runs the task until the PC is equal to the entered address. This function is not very fast because the task is running in single step mode and after each instruction the PC is compared with the address. * Trace out of OS runs the task until the PC is outside of the kickstart. This function is useful when you catch a task inside the OS and you want to get as fast as possible back to the program's code. It works similar as `Trace on address'. You shouldn't use this command if your task only runs in the kickstart. * (PC)++ jumps over the current command. Useful to jump over `Illegal' breakpoints used for debugging purposes in your program. * PC-2 subtracts 2 Bytes from the PC. * Write Nop overwrites the current command with a `Nop'. * Write Illegal overwrites the current command with an `Illegal'. * Run Task runs the task and stops only on exceptions. * Run Watched Task runs the task in trace mode and stops when a WatchPoint condition is true. If there are no watchpoints the command behaves like `Run Task'. * Run History Task runs the task in trace mode and saves the registers each step into the history stack. * Stop Task stops the task. * Send Signal sends a signal to the task. Default Signal is CTRL-C = 12 * Undo Level sets the undobuffer's depth. * Undo undos the last changes in the registerframe. * View Refresh activates and disactivates the copperlist refresh after each trace operation. This function is help for debugging programs which install own copperlists. * Show (EA) activates and disactivates the output of the address and address contents which are accessed by the current assembler command. * Symbol activates and disactivates the use of symbols in the `REGWindow'. * Delete Symbols erases all symbols of the task. * Copy Symbols can copy a symbol list of the task to a different task. This function is helpful if your task is started from another task and you want to keep the symbol list. * Load Symbols loads the symbols of a program where you can select an alternative process's segmentlist for calculating the symbol and debug informations. Normally you choose the same process but sometimes it's helpful to select a different process. For example, if the task you debug is created in a program, you have to choose the program's task to get the correct symbol addresses. * Set Hunklist sets a new segment list for the SourceWindow and some other hunk related functions. Because the position in the SourceWindow depends on the segments sometimes help to load new symbols and debug informations for this task. If you load an alternative Hunklist by selecting a custom task when you use `Load Symbols'. * Reset Hunklist removes the alternative hunklist. * Show Value shows the value of an argument. * Show Last Exception shows the last exception. * Open Task Window opens a window showing the task structure of the task. * Open System Window opens a window showing the ExecBase structure. * Open Proces Window opens a window showing the process structure of the process. * Open CLI Window opens a window showing the cli structure of the process. * Open Hunk Window opens a window showing the hunks of the process. * Open Symbol Window opens a window showing the symbols of the process. If you doubleclick on a symbol you get the hunk where the symbol is located. * Open Library Window opens a window showing the libraries. If you doubleclick on a library entry it opens a `FD:' window that shows all functions of the library when the library is defined in the `Barfly.FD' file. Furthermore if you doublelick on a function you have the choice to see the function in a `DissWindow' or the autodocs documentation. * Open Device Window opens a window showing the devices. * Open Resource Window opens a window showing the resources. * Open Port Window opens a window showing the public ports. * Open Resident Window opens a window showing the resident modules. * Open Interrupt Window opens a window showing the interrupts. * Open AutoDocs Window opens a filerequester which lets you choose the needed autodocs information of a library. This opens a window which shows all functions of the chosen autodoc file. If you now click on a function another window is opened showing the function documentation. * Open History Window opens a HistoryWindow showing the last saved registerframes of the undobuffer. The undobuffer is organized in a way to make the first entry become the last entry in the HistoryWindow. The HistoryWindow isn't updated automaticlly. * Stack Check controls the stack check. A warning is displayed if the register A7 points out of the stack bounds or points on an odd address. The Debugger only checks the task when the task gives back the control to the traproutine, so it's not possible to notice every stack problem. You should be aware that this function doesn't work with a WShell task because the `WShell' doesn't set the correct stack task values. * Find Task of address tries to find the task which belongs to the entered address. The command checks if the address is in the task, process, cli, mementry structure and the hunks. It's not safe to assume that the function can check all cases. * Load Binary loads a file with an optional length into a memory area. If you want the Debugger to automaticlly allocate the memory block you have to close the memory requester. * Save Binary saves a memory area into a file. * Freeze Task freezes a selectable task. When bdebug ends the frozen tasks are awakened again. * Warm up Task warms up a frozen task. * Kill Task kills a selectable task. You should know which task you can kill. For example, don't kill the input.device or your filesystem. * Show Task shows the task structure of a selectable task. * Show Prozess shows the process structure of a selectable process. * Show CLI shows the cli structure of a selectable process. * Show Hunk shows the hunks of a selectable process. * Send Task Signal sends a signal to a selectable task. * Set Task Priority sets a priority of a selectable task. * Refresh Code Cache refreshes the Code Cache. * Refresh Data Cache refreshes the Data Cache Usage of BDebug *************** The FPU Window ============== FPU Window ---------- The FPU Window shows the FPU register FP0 to FP7 in the 96Bit Extended format and the registers FPCR, FPSR and FPIAR in hexadecimal. You can only open this window if a FPU is available. Register Window Layout FP0=FloatingPoint FP1=FloatingPoint FP2=FloatingPoint FP3=FloatingPoint FP4=FloatingPoint FP5=FloatingPoint FP6=FloatingPoint FP7=FloatingPoint FPCR=xxxxxxxx FPSR=xxxxxxxx FPIAR=xxxxxxxx Local Menus ----------- * Close Window closes the window. Usage of BDebug *************** The Disassembler Window ======================= Disassembler window ------------------- The `DissWindow' shows the memory contents in assembler mnemonics. The address of the window's view can be absolute or relative. In absolut mode, the window shows the contents of a fixed address, indicated by the window title `No Link'. In relative mode the window is connected to a register causing the window to update when the the register value changes. This is indicated by the window title which reads `Link to *', where * represents the register name. The PC is shown by the colour pen 2. If the linked register value is outside of the window's view area the whole contents of the window will be refreshed. You can change size of the window and scroll through the memory area by using the cursors. In the title you see an ID-String with the format `\#x.y' where `X' represents the `REGWindow' number and `y' the number of the `MemWindow'. Doubleclicking a line of the window sets or remove a breakpoint. You can disable this function in the configuration. Local Menus ----------- * Close Window closes the window. * Shrink Window shrinks the window. * Expand Window expands the window to screen size. * Link to Register links the window with a register. If you enter the string `NO' it switches to absolute mode. * Change address changes the view address of the window. * Clear address resets the view address of the window. * Refresh Window refreshes the window. * .W Branches activates and disactivates the output of the old branch width size. * Neg. Offsets activates and disactivates the output of negative values in indirect address modes with offset. * Neg. Data activates and disactivates the output of negative values in direct address mode. * Opcode Data activates and disactivates the additional output of the command bytes. * Auto Refresh activates and disactivates the global refresh of the window after each step. * Symbols activates and disactivates the symbol output in the window. * Show Lib Call activates and disactivates the symbolic output of library functions so all library functions that are defined in the configuration file `<BarflyPath>/Barfly/BARFLY.FD' are recognized. * Guess Lib Call activates and disactivates the guessing of function call names. This only works in connection with the option `Show Lib Call'. Unfortunately you can't expect that the function names always fit because the library base register A6 can change until the program counter meets the function. * Mark Block End activates and deactivates marking after the instruction `JMP', `BRA', `RTS', `RTE', `RTD' and `RTR' to make program blocks more visible. * Set/Clear Breakpoint sets/removes a breakpoint on the first entry in the window. Breakpoints are shown by changing the pen from colour 1 to colour 3 and the char `>' at the beginning of a line. * Pick/Clear Breakpoint sets/removes a breakpoint through a symbol list. * Disassemble to File disassembles a memory area into a file. Usage of BDebug *************** The Memory Window ================= Memory window ------------- The `MemWindow' shows the memory contents hexadecimal and in ascii.You can change size of the window and scroll through the memory area whith the cursor keys. In the title you see an ID-String with the format `\#x.y' where `X' represents the `REGWindow' number and `y' the number of the `MemWindow'. Local Menus ----------- * Close Window closes the window. * Shrink Window shrinks the window. * Expand Window expands the window to screen size. * Link to Register links the window with a register. If you enter the string `NO' it switches to absolute mode. * Change address changes the view address of the window. * Clear address resets the view address of the window. * Refresh Window refreshes the window. * Memory Offset Step defines the data format in the window. The following options can be selected: `None', `Byte', `Word' and `Long'. * Edit activates edit mode of the `MemWindow'. In edit mode you can switch between hex and ascii input by the key `RETURN'. With `ESC' you can leave edit mode. Only the cursor right and left are changed to the normal. With these both keys you can access each Byte. In edit mode you can't change the size of the window. * Copy copies a memory area into another memory area. The function uses `CopyMem' so it doesn't handle memory areas that overlap. * Fill fills a memory area with a value of a certain data-width. * Compare compares a memory area with another memory area. * Search searches a value of a certain data-width in a memory area. If the value is found, the address and the value are displayed and you can goon with `Search Next' to find the next address. * Search Next Search the next value. Look at `Search' * Pred sets the address of the window tn the preceding entry of the list. If the node points on an odd, illegal or NULL address the command has no effect. The next node is equal to `LN_PRED', the second longword of the memory view. * Succ sets the address of the window to the next entry of the list. If the node points on an odd, illegal or NULL address the command has no effect. The next node is equal to `LN_SUCC', the first longword of the memory view. Usage of BDebug *************** The Copper Window ================= CopperWindow ------------ The `CopperWindow' shows the memory contents as copper commands. The window is sizeable, and you can use the cursor keys to scroll through the memry area. In the title you see an ID-String with the format `\#x.y' where `X' represents the `REGWindow' number and `y' the number of the `CoppWindow'. Local Menus ----------- * Close Window closes the window. * Shrink Window shrinks the window. * Expand Window expands the window to screen size. * Link to Register links the window with a register. If you enter the string `NO' it switches to absolute mode. * Change address changes the view address of the window. * Clear address resets the view address of the window. * Refresh Window refreshes the window. * Goto Into List sets the window list on the standard copperlist `GfxBase->gb_copinit'. Usage of BDebug *************** The Structure Window ==================== StructWindow ------------ opens window which can be connected with a structure. You can use new structure entries by expanding the the `<BarflyPath>/Barfly/Barfly.Include' file or loading a new custom file. By a doubleclick on a structure window entry you can cause several actions depending on the datatype. Every datatype is connected with an action that is normally started automaticlly. With the configuration command `NoAutoStructAction' you can change this behaviour so that an action type requester is opened. The following datatypes are available. * APTR opens a MemWindow. * CSTR shows a string. * BPTR opens a MemWindow at the address BPTR*4. * BSTR shows a string at the address (BPTR*4)+1 * CPTR opens a DissWindow. * FPTR opens a DissWindow. * BYTE doesn't cause an action. * WORD doesn't cause an action. * LONG doesn't cause an action. * FLOAT doesn't cause an action. * DOUBLE doesn't cause an action. * EXTENDED doesn't cause an action. * RPTR doesn't cause an action. The following action types are available. * MemWindow opens a MemWindow. * DissWindow opens a DissWindow. * CoppWindow opens a CoppWindow. * StructWindow opens a StructWindow. * NewStruct sets a new structure. Structure Macro Fileformat -------------------------- This section describes the file format of structure macros. In the beginning you define the root directory entries with the first Macro `Menudir'. The first parameter is the name of the entry, then the address of the parent directory and then the address of the subdirectory. In the root directory the parent address is obviously NULL. The last entry of the directory is defined by the Macro `MENUDIREND'. `Label ListViewMacro Link' RootDir: . . MENUDIR exec,0,Exec_Dir . . MYCUSTOMENTRY: MENUDIREND CUSTOM,0,0 The design of a subdirectory only differs from the root directory entries only by a parent directory address. `Label ListViewMacro Link' Exec_Dir: . . MENUDIR nodes.i,RootDir,Nodes_Dir . . MENUDIREND tasks.i,RootDir,Tasks_Dir to define the structure directory entries you have to use `MENUITEM' and `MENUITEMEND'. The first parameter in the Item Macros is the name of the entry and also the name of the structure. The second parameter defines the address of the parent directory. `Label ListViewMacro Link' . . Nodes_Dir: MENUITEM LN,Exec_Dir . . MENUITEMEND To define a structure you can use the normal assembler syntax that you probably have to adjust to your custom needs. For example you can tell BDebug more informations about the datatype an entry represents. By redefining `APTR' to a `CSTR' you can tell the Debugger that the entry is a stringpointer. Another example is defining that `APTR' points to a structure by `APTR LN_SUCC,Node'. `Label IncludeTypeMacro Name,Link' LN_Struct: STRUCTUREB LN,0 APTR LN_SUCC,LN APTR LN_PRED,LN UBYTE LN_TYPE BYTE LN_PRI CCSTR LN_NAME LABEL LN_SIZE Local Menus ----------- * Close Window closes the window. * Shrink Window shrinks the window. * Expand Window expands the window to screen size. * Link to Register links the window with a register. If you enter the string `NO' it switches to absolute mode. * Change address changes the view address of the window. * Clear address resets the view address of the window. * Refresh Window refreshes the window. * Load Custom Struct loads additional structure files. The new structure entries are placed in the `CUSTOM' directory. The format of custom structure files is equal to the file `BARFLY.INCLUDE'. * Select Structure opens the structure include directory requester where you can select the needed structure. The parent gadget is placed in the upper border. * Goto Sysbase... sets the window address on the ExecBase. * Goto Gfxbase... sets the window address on the GFXBase. * Save Window.... saves the contents of the window in a file. * Full Address this switch decides whether the `StructWindow' also shows the address of the entries. * Offset Address this switch decides wheather the `StructWindow' also shows the offset of the entries. * Pred sets the address of the window on the preceding entry of the list. If the node points on an odd, illegal or NULL address the command has no effect. The next node is equal to `LN_PRED', the second longword of the memory view. * Succ sets the address of the window on the next entry of the list. If the node points on an odd, illegal or NULL address the command has no effect. The next node is equal to `LN_SUCC', the first longword of the memory view. Usage of BDebug *************** The Structure Window ==================== Source window ------------- The `SourceWindow' shows the source line that belongs to the window address. If the program file doesn't have the needed debug informations the `Source window' can't be opened. If the address points to an area with no relevant debug information, for example the Kickstart or beyond the program hunks, all you see is a small message. Local Menus ----------- * Close Window closes the window. * Shrink Window shrinks the window. * Expand Window expands the window to screen size. * Link to Register links the window with a register. If you enter the string `NO' it switches to absolute mode. * Change address changes the view address of the window. * Clear address resets the view address of the window. * Refresh Window refreshes the window. * Set Breakpoint sets a breakpoint on the active line. * Show HunkInfo shows the hunk of the current source line. Usage of BDebug *************** The Snoop Window ================ Snoop Window ------------ The `SnoopWindow' snoops the task's allocations. Local Menus ----------- * Close Window closes the window. * Shrink Window shrinks the window. * Expand Window expands the window to screen size. * Refresh Window refreshes the window. * Auto Refresh activates/deactivates display refresh by an allocation. * Snoop Memory activates/deactivates snooping. * Snoop Mask sets the allocation filter mask. If the Mask is set to , only allocations with the size 20 are recorded. Default -1. * Snoop Max Entries sets the maximal recordable snoop entries. Usage of BDebug *************** The Breakpoint Window ===================== Breakpoint window ----------------- The `BreakWindow' handles all breakpoints and contains the functions that are needed with breakpoints. In general, breakpoints are addresses in the program where the task should be stopped. The breakpoints are handled globally so they aren't deleted when close the window. Local Menus ----------- * Toggle activates and deactivates all breakpoints. * All selects all breakpoints. * Clear unselects all breakpoints. * On activates all selected breakpoints. * Off deactivates all selected breakpoints. * Hit defines how often a breakpoint must be encountered before the program stops. Default is 1. * ? shows the hunk where the breakpoint is located and also shows if the breakpoint is equal to a symbol. * Input sets and removes a breakpoint. * Pick sets and removes a breakpoint using the symbol list. * Delete removes all selected breakpoint. * Goto opens a DissWindow for each selected breakpoint. * Run runs the program until the PC encounters a selected breakpoint. Usage of BDebug *************** The Watchpoint Window ===================== Watchpoint window ----------------- The `Watchwindow' allows to set breakpoints that aren't depending on a certain PC address but on other conditions. Every watchpoint has a condition, data width and can have one of 3 different watchpoints states. The `Memory' watchpoint compares the saved contents of the address with the current contents and dependent on the condition the program is stopped or not. The `Register' watchpoint compares the saved contents of a register with the current contents and dependent on the condition the program is stopped or not. The `Argument' watchpoint compares the saved value of an argument with the current contents and dependent on the condition the program is stopped or not. The last watchpoint type is the most powerful because it can simulate the first two types with the cost of a slowdown. The use of watchpoints is very time consuming because the whole program is run in single step mode. In order to use watchpoi you have to run the task with `Run Watched Task'. If an error occurs in the exception handler during the evaluation of a dynamic arugment, the screen flashes. Local Menus ----------- * Toggle activates and deactivates all watchpoints. * All selects all watchpoints. * Clear unselects all watchpoints. * On activates all selected watchpoints. * Off deactivates all selected watchpoints. * Add opens a requester where the parameters for a watchpoint have to be adjusted and adds the new watchpoint to the list. You can change the parameters simply by doubeclicking on a watchpoint. * Refresh refreshes the watchpoint arguments. * Check checks all selected watchpoints. * Delete removes all selected watchpoints. Usage of BDebug *************** The Checksum Window =================== Checksum Window --------------- The `ChecksumWindow' controls all checksum areas that are controlled each time the task stops. Helpful to find illegal random writes bugs. The checkpoints are controled globally so they aren't deleted when you close the window. Local Menus ----------- * Toggle activates or deactivates all checksum areas. * All selects all checksum areas. * Clear unselects all checksum areas. * On activates all selected checksum areas. * Off deactivates all selected checksum areas. * Address adds a checksum area into the list. * Hunk adds a hunk of the current process into the checksum area list. * Task adds a hunk of selectable process into the checksum area list. * Refresh calculate a new checksum for all selected areas. * Delete removes all selected checksum areas. * Check checks all areas for checksum errors. Usage of BDebug *************** Requester Arguments =================== Argument Structur ----------------- An argument can use absolut values, symbols and registers as operands and the operators +,-,*,/,|,!,&,<<,>>,~. Additionally to the normal symbols, some special symbols are available. * By {Argument}.[b,w,l] you can read from a memory address, that is defined by the argument. An error is indicated if you specify an illegal address with isn't defined in the legal memory space. * \#d? represents the address of the `Disswindows' with the ID ? * \#m? represents the address of the `Memwindows' with the ID ? * \#c? represents the address of the `Coppwindows' with the ID ? * \#h? represents the address of the hunk ?. Helpful for Enforcer and CyberGuard Hunk:Offset output * \#ea? represents the address of the EA with the number ?. Check Register Window. * \#em? represents the contents where the address EA number ? points to. Check Register Window. If the address EA is illegal an error is shown. * \#ls represents the start address of a loaded binary file. * \#le represents the end address of a loaded binary file. * \#ll represents the length address of a loaded binary file. * \#p represents the start address of the programs. Only true for a loaded program. If you have the following Enforcer or CyberGuard hit output `Hunk 0:$11c' you can calculate the address by entering the argument #h0+$11c. Usage of BDebug *************** Technical Informations ====================== Exceptions ---------- The Debugger can catch all exceptions if the system is still working. If an exception is caused, the traphandler catches the exception and tells the Debugger what went on so it can react on the exception. If the exception wasn't caused by the Debugger, the type and the possible reason for the exception is shown. The `Return-address' of the debugged task points on an internal `ILLEGAL'. If the processor encounters this `ILLEGAL' the task is closed and all windows are removed. You shouldn't step over this `ILLEGAL' because it increases the possibility of a system crash. If a task is caught by `Debug Next Task' and notices a custom `finalPC' routine in the `Addtask()' function, the `Return-address' isn't set on the internal `ILLEGAL' cleanup function because the `finalPC' pointer is sometimes used for parsing an argument. In this case the Debugger notices that the task ends by using the `RemTask()' function. If the task changes the `Return-address' the Debugger tries to determine the taskend by `RemTask'. Exception Handler ----------------- Every task has a pointer in its task structure that points to its exception handler, named `TC_TRAPCODE'. When you load a program through the Debugger, the Debugger's standard exception handler is installed. It works basicly the same for `Catch Next Task' and `Debug Task'. The only difference is the `Debug Crashed Task' mode where the Debugger sets a special catch exception handler in each task that isn't controlled by the Debugger. When an exception occurs and the Debugger's exception handler is called the Debugger first checks if it knows the task that caused it. If this is not the case something seriously is broken and the Deadend Alert 35000000 will be popped up. If all goes well the registers are saved, the Debugger task gets a message and the exception handler waits for a message by the Debugger to go on. When the Debugger gets the message it causes the appropriate function. For example refreshing the windows. If the Debugger gets a step command it sends the exception handler the appropriate message and the handler does a step. Debug Informations ------------------ Currently the following formats are supported. * BASM Specialformat This format allows the Debugger to decide if the code is in the Mainpart, Includes or in a Macro. * SAS D1 This format only allows a Source-Code connection. It doesn't support local variables, Structures and Macros. * GCC STABS This format is very powerful and offers all a source-level Debugger needs. Unfortunately the Debugger only supports a simple Source-Code connection at the moment. It's planned to support more in the future. GCC Compiler and BDebug ----------------------- Unfortunately you can't debug programs that are using the current `ixemul.library' because in Openlibrary() initroutine the Task field TC_TRAPCODE is changed. Hopefully there'll be soon an `ixemul.library' available that doesn't change the traphandler. If you're using GCC with the link lib `gerlib' that is available on Aminet FTP Servers you shouldn't experience any problems with BDebug. Usage of BDebug *************** Configuration ============= The default configuration file is named `BDEBUG.Config' and is located in the directory `<BarflyPath>/' or `s:Barfly/'. Obviously it's not optimal to be forced to use the same config file for different programs. Therefore you can also specify a local config file with program name and the suffix `.BDebug'. ToolTypes --------- The following tooltypes are supported to activate the basic functions of the commandwindow. * CatchNextTask * CatchCrashedTask * CatchEnforcerHit Barfly.FD --------- Following steps are necessary to create a new `Barfly.FD' file with new library functions. First the assign `FD:' must point to the directory containing the FD files that should be included in the new `Barfly.FD'. Make sure that every FD file contains the Library, Resource and Device name in the first line in following format: `* "foobar.libary"'. If necessary, add the name yourself, so that a correct FD database can be build up. If you're more experienced with FD files, you can yourself add new entries to the `Barfly.FD' file. The layout of the `Barfly.FD' file pretty obvious. Configuration Commands ---------------------- Register window --------------- RegWindow=x/y/width/height/register ................................... This command defines the position of a `REGWindow'. RegFlags=flag[|flags...] ........................ This command defines certain flags in the `REGWindows'. * AUTOVIEWREFRESH * SYMBOLS * STACKCHECK * NOBIGVIEW FPU Window ---------- FpuWindow=x/y/width/height/register ................................... This command defines the position of a `FPUWindow'. OpenFPUWindows=Count .................... This command tells the Debugger to open a `FPUWindow'. Disassembler window ------------------- DissWindow=x/y/width/height/register .................................... This command defines the position,the dimension and linked register of the `DissWindow'. Example: DISSWINDOW=0/0/300/100/PC DissFlags=flag[|flags...] ......................... This command defines certain flags in the `DISSWindow'. * AUTOREFRESH * SHOWLIB * GUESSLIB * SHOWEA * WORDBRANCHES * NEGOFFSETS * NEGDI * OPCODEDATA * BLANKAFTERJMPBRA OpenDissWindows=Count ..................... This command tells the Debugger to open a number of `DissWindows'. Memory window ------------- MemWindow=x/y/width/height/register ................................... This command defines the position, the dimension and linked register of the `MemWindow'. Example: MEMWINDOW=0/0/300/100/A0 OpenMemWindows=Count .................... This command tells the Debugger to open a number of `MemWindows'. MemoryOffsetStep=Count ...................... This command defines the Offset-Step of the `MemWindows'. * 0 no Space * 1 Space after each Byte. * 2 Space after each Word. * 4 Space after each Longword. Copper window ------------- CoppWindow=x/y/width/height/register .................................... This command defines the position, the dimension and linked register of the `CoppWindow'. Example: COPPWINDOW=0/0/300/100/A0 OpenCoppWindows=Count ..................... This command tells the Debugger to open a number of `CoppWindows'. StructWindow ------------ StructWindow=x/y/width/height/register ...................................... This command defines the position, the dimension and linked register of the `StructWindow'. Example: StructWINDOW=0/0/300/100/A0 StructFlags=flag[|flags...] ........................... This command defines certain flags for the `StructWindow' * AUTOREFRESH * ADDRESSFORMAT * OFFSETFORMAT * NEWWINDOW OpenStructWindows=Count ....................... This command tells the Debugger to open a number of `StructWindows'. Source window ------------- SourceWindow=x/y/width/height/register ...................................... This command defines the position, the dimension and linked register of the `SourceWindow'. Example: SOURCEWINDOW=0/0/300/100/A0 OpenSourceWindows=Count ....................... This command tells the Debugger to open a number of `SourceWindows'. Breakpoint window ----------------- BreakWindow=x/y/width/height ............................ This command defines the position of a `BreakPointWindow'. OpenBreakWindows=Count ...................... This command tells the Debugger to open a `BreakPointWindow'. Watchpoint window ----------------- WatchWindow=x/y/width/height ............................ This command defines the position of a `WatchpoinzWindow'. OpenWatchWindows=Count ...................... This command tells the Debugger to open a `WatchPointWindow'. Checksum window --------------- ChecksumWindow=x/y/width/height ............................... This command defines the position of a `ChecksumWindow'. ChecksumWindows=Count ..................... This command tells the Debugger to open a `ChecksumWindow'. SnoopMemory window ------------------ SnoopMemWindow=x/y/width/height/register ........................................ This command defines the position of a `SnoopMemWindow'. OpenSnoopMemWindow=Count ........................ This command tells the Debugger to open a `SnoopMemWindow'. SnoopMask=Mask .............. This command defines the snoop mask. The mask defines the length of memory blocks that should be recorded. Default is -1 so everything is recorded. SnoopMax=Count .............. This command defines the count of snoop entries. Default is 100. HistoryEntrys=Count ................... This command defines the count of history entries. Default is 5. Information Windows ------------------- GlobalViewWindow=x/y/width/height ................................. This command defines the position and dimensions of a standard information window. For example the Library Window belongs to this group. Example: GLOBALVIEWWINDOW=0/0/300/100 GLOBALVIEWWINDOW=0/200/300/100 Other Windows ------------- CommandWindow=x/y/width/height .............................. This command defines the position of the small `CommandWindow'. This command has no function in local configuration files. FileShell=<Window Specifikation> ................................ This command defines the shell that is opened with the loaded program. You should always open the shell on the Debugger's Public Screen. The shell parameters are the same you know from the CLI. Misc ---- TaskStack=Count ............... This command defines the stack of the loaded program. Defaullt are 4096 Bytes. Priority=Count .............. This command defines the Debugger's priority. SetBreak=Argument ................. This command can be used to define a list of breakpoints that are set before the program is started. This is useful to pass the module `Main.c' for example. If no breakpoints are defined or if a parsing problem occurs the standard breakpoint ( the first program instruction ) is set. * SETBREAK=_main ; SAS C Main Program Start * SETBREAK=_main ; GCC C Main Program Start * SETBREAK=@main ; DICE C Main Program Start * SETBREAK=! ; Programstart(Default) ClickBreak=State ................ This command can be used to define the action of the DissWindow on a doubleclick. `State=0' No Action(Default). `State=1' Set/Clear Breakpoint and pop up a Requester for a Set. `State=2' Set/Clear Breakpoint. ShowMem=Start:End ................. defines the address areas that are legal to the Debugger so you can look at address areas that are not in the memorylist or in the rom. Illegal address areas are shown with `*' in the windows. You should `never' define the custom chip areas as legal because a read access on a writeonly register can cause a deadly crash. Example: SHOWMEM=$e80000:$f00000 defines the Zorro 2 area as free. By this command you can overrule the internal Enforcer or CyberGuard legal memory areas so you should be free of hits. DefCommand=Key,Qualifier[|Qualifier...],Function ................................................ This commands allows to connect menu functions with key sequences. Because of the object-oriented concept of the Debugger that allows multiple instances of objects it's not easy to decide what object is meant. Therefore if the object is active it's used and if no object of this type is active the first entry the object-type list is used. As the key parameter every Rawkey can be used with the exception of `TAB' and the functionkeys that are used internally. The key is searched first in the local and then in the global configuration. As qualifiers you can use the following keys. * LSHIFT * RSHIFT * CAPSLOCK * CTRL * LALT * RALT * LCOMMAND * RCOMMAND Bespiel: DEFCOMMAND=$15,CTRL,"Step 1 Position" Defines CTRL-Z as Step 1 Position AutoDocDir=<Path> ................. This command sets the path for the autodocs directory. AutoDocAlias=Library,File ......................... This command sets an alias for Libraries, Devices or Resources to define the connected Autodocs file. There's no other way because it's impossible to build the autodocs file by knowing the library name. ArexxPath=<rx-path> ................... This command sets the Arexx-Script Start-Command. In a normal system the path should be <sys:Rexxc/rx>. ArexxInput=<File> ................. This command sets the Arexx-Command Input-File. If you don't specify the file, NIL: is used. ArexxOutput=<File> .................. This command sets the Arexx-Command Output-File. If you don't specify the file, NIL: is used. ArexxCommand=[1...10],<Pfad> ............................ This command sets the 10 entries in the Arexx-Menu. You specify the number and the path of the Arexx-Script. ArexxCommand=1,"Rexx:Example.rexx" ExecuteCommand=<File> ..................... this command can set up a list of programs that should be run before the debugged program's task is started. This parameter only works with programs which are loaded by the Debugger. You also have to make sure that the loaded programs have to `end' otherwise the task can't be started. For example you could use this command to set breakpoints with Arexx-Scripts. LoadInclude ........... tells the Debugger to load the structure information file `Barfly.Include'. AddStructFile=Filename ...................... tells the Debugger to load a custom structure information file and adds it into the CUSTOM/ subtree. ClicktoFront ............ activates the Debugger's own `ClicktoFront' handler. This function should only be used if you don't use an own `Commodity' for this task. CenterWindow ............ activates centering mode for all stringrequester windows. ScreenInFront ............. activates `ScreenToFront' mode that pops the screen to front after every trace operation. OpenScreen[=width,height,depth,mode] .................................... tells the Debugger to open its own screen. If you don't enter dimension parameters the wb screen is cloned. For the mode string you string you can see in `Prefs/ScreenMode' requester. This command has no function in local configuration files. OPENSCREEN=1448,560,2,PAL:HighRes Interlace OpenPubScreen=Name .................. tells the Debugger to open on the Pubscreen with the specified name. This command has no function in local configuration files. ScreenFont=fontname/Height .......................... defines a font for a Debugger screen. This command has no function in local configuration files. QuietException=Exception Nummer ............................... masks off certain exceptions for the exception requester so that only a `DisplayBeep' is caused instead of a textrequest. With the value -1 you can mask off every exception and for example with the value 4 you mask off the `Illegal' exception. DisableXPointer ............... deactivate the Wait-Pointer. TraceBreak .......... tells the Debugger to use breakpoints in the `Subroutine' Traces instead of single steps. The advantage is a speed up and the disadvantage is that you can cause crashes while you step through resident/reentry code. CrashedTask ........... activates `CatchCrashedTask' mode. CatchEnforcerHit ................ activates `CatchEnforcerHit' mode. DoNotCacheFullFile .................. tells the Debugger not to cache program files while reading the Symbol/Debug informations to save memory. Obviously the parsing speed will decrease. DoNotPopPathRequest ................... tells the Debugger to ignore errors from opening source files and not to open a path requester. NoAutoStructAction .................. tells the Debugger to open a type-requester by an action in the Structure Window. NoBreakpointErrors .................. tells the Debugger to ignore SETBREAK= errors that cause the Debugger to always set an error on the program start. Usage of BDebug *************** Arexx ===== Commands -------- SIMPLEREQUEST " ............... * RC: - * result: 'OK' TWOGADREQUEST " ............... * RC: - * result: 'OK','FALSE' TRIGADREQUEST " ............... * RC: - * result: 'OK','FALSE','RESUME' NEXT_ROOTWINDOW ............... * RC: - * result: 'OK','FALSE' NEXT_SUBWINDOW .............. * RC: - * result: 'OK','FALSE' FIRST_DISSWINDOW ................ * RC: - * result: 'OK','FALSE' FIRST_MEMWINDOW ............... * RC: - * result: 'OK','FALSE' FIRST_COPPWINDOW ................ * RC: - * result: 'OK','FALSE' FIRST_FPUWINDOW ............... * RC: - * result: 'OK','FALSE' FIRST_BREAKPOINTWINDOW ...................... * RC: - * result: 'OK','FALSE' FIRST_STRUCTWINDOW .................. * RC: - * result: 'OK','FALSE' FIRST_SOURCEWINDOW .................. * RC: - * result: 'OK','FALSE' FIRST_SNOOPWINDOW ................. * RC: - * result: 'OK','FALSE' FIRST_WATCHWINDOW ................. * RC: - * result: 'OK','FALSE' ACTIVATE_ROOTWINDOW ................... * RC: - * result: 'OK','FALSE' ACTIVATE_SUBWINDOW .................. * RC: - * result: 'OK','FALSE' OPEN_DISSWINDOW '@REG' | 'Argument' ................................... * RC: - * result: - OPEN_MEMWINDOW '@REG' | 'Argument' .................................. * RC: - * result: - OPEN_COPPWINDOW '@REG' | 'Argument' ................................... * RC: - * result: - OPEN_SOURCEWINDOW '@REG' | 'Argument' ..................................... * RC: - * result: - OPEN_STRUCTWINDOW '@REG' | 'Argument' ..................................... * RC: - * result: - OPEN_BREAKPOINTWINDOW ..................... * RC: - * result: - OPEN_FPUWINDOW .............. * RC: - * result: - OPEN_SNOOPMEMORYWINDOW ...................... * RC: - * result: - DOMENU 'Menu-String' .................... * RC: - * result: - SET_BREAKPOINT 'Argument' ......................... * RC: - * result: 'OK','FALSE' CLEAR_ICACHE 'Address,Length' ............................. * RC: - * result: 'OK' CLEAR_ICACHE 'Address,Length' ............................. * RC: - * result: 'OK' GOTO_ADDRESS 'Address' ...................... * RC: - * result: 'OK','FALSE' CLEAR_ADDRESS ............. * RC: - * result: 'OK','FALSE' LINK_REGISTER 'Register' ........................ * RC: - * result: 'OK','FALSE' SET_REGISTER 'Register,Value' ............................. * RC: - * result: 'OK','FALSE' Read_Byte 'Address' ................... RC : 0=Ok Result: Result-String Read_Word 'Address' ................... RC : 0=Ok Result: Result-String Read_Long 'Address' ................... RC : 0=Ok Result: Result-String Write_Byte 'Address,Value' .......................... RC : 0=Ok Write_Word 'Address,Value' .......................... RC : 0=Ok Write_Long 'Address,Value' .......................... RC : 0=Ok ASL_FileRequester_Save ...................... RC : 0=Ok Result: Filepath-String ASL_FileRequester_Load ...................... RC : 0=Ok Result: Filepath-String IS_ADDRESS_LEGAL 'Address' .......................... * RC: - * result: 'OK','FALSE' LOAD_BINARY 'Name,Destination,Length' ..................................... SAVE_BINARY 'Name,Source,Length' ................................ Usage of BDebug *************** How to use BDebug ? =================== Trouble Shooting ---------------- First you should be sure that all necessary configurations file have been installed because without `Barfly.FD' file you don't see any function names in the disassembler window; and without `Barfly.Include' the `StructWindow' is unusable. Are these preconditions fullfilled you should analyse the problem and anticipate how the Debugger can be used. Because the debugging of programs depends heavily on the situation i can only list some general points. The reality probably looks different... as always:-) Point of departure: * Program from the CLI * 1 Task The program can be started by `bdebug Program [Argument]' or can be loaded the command window `Debug File'. By this method all symbol and debug files are loaded. If the Debugger can't find a source file you can add additional paths if you haven't disabled this function. The standard breakpoint is the first command in the program. Sometimes this is inconvenient, and you may want to set a different start breakpoint for example to jump over the CStartup code or to set it tn an important program position. * Creates further Tasks In this case you should be sure how you want to catch the next Task. You could catch the Task by `Next Task' or compile the program with an `illegal' in the task and catch it with `Crashed Task'. After you caught the task you probably would like to use see symbols and debug source. These informations can be loaded afterwards by using `Load Symbols'. * Program from the WB In this case you should use `Next Task' and then catch the task `WBL'. Afterwards you have to activate `Next Task' again and run `WBL'. You could also use the `illegal' strategy. After the right task was caught you can load the symbols again. * Is it a Handler, Filesystem or something similar. In this case you should use the `illegal' strategy and catch the task by `Crashed Task'. An alternative method would be to catch the waiting task with `Debug Task' and wait as long as the task gets woken up by a signal. After you've taken the first hurdles in taking control over the task, you should think about how the problem looks like and where it could be located. Problem Type: * Enforcer/CyberGuard If an Enforcer or CyberGuard hit is caused, these programs output the hit's program address and most of the time the hunk offset as well. You can now directly jump to the address by entering the address in the DissWindow by `Change Address' or you open a HunkWindow, doubleclick on the hunk where the hit is located and then enter the offset returned by Enforcer or CyberGuard. The Debugger itself can also automaticlly stop a debugged program if a hit happens. * Mungwall These hits aren't as easy to find as Enforcer or CyberGuard hits because Mungwall hits aren't shown when the problem happened but only after a `FreeMem'. In this case you should remember the memoryblock where it happened and determine where the responsible AllocMem is located in the program, so you get an overview in what area the problem is caused. Now you should open a MemWindow that points to this certain memory area and step through the program and look if something changes the mungwall borders in the MemWindow. Mungwall borders are before and after the allocated memory area. If you're more experienced you could also use the WatchpointWindow and set a watchpoint on the certain memory block. * Crash If it's just an ordinary crash the error should be pretty easy to find by single stepping through the responsible code area. If it's a random crash you should try `Crashed Task' and hope that the task can be caught. After the task got caught you should check the instructions that caused the crash. If the PC points to data fields that don't look like real code the PC is probably set wrong by a stack cleanup error. In this case you should check if the next addresses on the stack point to legal program code. * Side effects and mysterious bugs These bugs are the worst to find and there's no general strategy how to find them. In such cases only intuition and patience can help. Problems that can happen ------------------------ * Why does the Debugger react so slow on keyboard commands that control the tracing ? This happens if you debug a task with a higher priority than the Debugger's priority. For example. DOS-Handler. Workaround is to increase the priority of the Debugger. * Why do filerequesters block the Debugger. This happens when you debug a handler, because the filerequester normally tests every handler with a IsFilesystem(). When you debug a handler it can't reply the IsFilesystem packet and therefore the filerequester is busy. * If you debug the following program on a 68040 with 68040.library or 68060.library the instruction `rts' is run if you cause a Single-Step on the instruction `fetox'. Because this command isn't implemented in the 68040 and 68060 it has to be emulated. It seems to forget the tracebit. mc68040 fmove.x #1.3,fp0 fmove.x #255,fp1 fetox.x fp0,fp1 rts * If a program doesn't return from a DOS function, check whether you accidently entered a char in shell window. BAsm 1.0 A cli/arexx controled Assembler Copyright (c) 1989-94 Ralph Schmidt - Shareware - The Assembler ************* The assembler understands the commands and addressmodes from the 68000 through the 68060 and both the Floating-Point Units, 68881 and 68882. It supports only the 68851 MMU commands, which are also supported by the 68030. The assembler achieves it's speed by translating the source in a single pass, followed by a backpatch phase which corrects all unresolved references. Syntax ====== Comments -------- A comment can start in several different ways. In a pure comment it starts either with a ; or *. A comment can only be started after an assembler command or symbol with a ; if an assembler command or symbol exists within that line. ;A comment *A comment move.l a0,a0 ;A comment Opcode/Instructions arrangement ------------------------------- `[label[:]] [opcode] [operand[, operand[, operand...]]]]]' * Opcodes An Opcode can be a Motorola Mnemonic, an assembler command, or a Macro call. * Operations In a Motorola mnemonic operands are based on legal addressmoes; in assembler privat instructions the parameters depend on the instruction. Symbol structure ---------------- A symbol can represent the following types: * Value * Program Counter * Register * Register List * Macro Symbols can only be defined once. The exceptions are local labels and symbols defined by Set. Structure rules for Symbols. * The first letter of a symbol can be one of the following: a...z, A...Z,_, @,. and \. * From the second letter on, the symbol can contain the following letters: a...z, A...Z, 0...9, _,@ and .. * If a symbol consists of only numbers and ends with $, then it is a numerical local label. * A symbol is ended by an illegal letter. * If a symbol begins with a . or \, then it is a local label. * A macro symbol should not contain a .. * To avoid a conflict, a symbol should not end with .b, .w or .l. ThisIsALabel ;That is a normal label ThisIsALabel1.loop: ;That is a normal label This@_Is_@A_@Label 1$: ;That is a numerical local label .ThisIsALabel ;That is a local label .ThisIsALael1: ;That is a local label \ThisIsALabel.loop: ;That is a local label ThisIsASymbol=10 ThisIsASymbol = 10 ThisIsASymbol equ 10 The relative label ------------------ This symbol represents an offset to the start of a program. label label: label nop label: nop The local Label --------------- A local label is only valid between two normal relative labels, thus you cannot reference local labels outside of that scope. Otherwise it works similar as a normal label. There are 2 different prefixes that introduce a local label: . and \ that define 2 different local labels. A special case is the Backward Reference Label that is introduced with ... It doesn't depend on a certain define area between normal labels thus you can only access the symbol if it were defined earlier. ..: label_0: .local: bra.s .local label_1: .local: bra.s \local \local: nop label_end: ..: dbra d0,.. ..: dbra d1,.. ..Hello: dbra d1,..Hello The local numerical label ------------------------- Addionally to the non-numerical local label there are also the numerical labels which are based of 4 digits with the postfix $. BASM handles the number as a hash key with the consequence that there's no difference between 001$ and 1$. label_0: 123$: nop label_1: 123$: nop The absolute Symbol ------------------- The absolute symbol is defined by a direct value initializing that is initiated by =, equ or set. If you define a symbol by set you can change it as often as needed. value1=2 value2 equ value1*2 value3 set value2 The Register Symbol ------------------- The register symbol is defined by `equr' or `fequr' that is used for FPU registers. Ptr equr a1 PI fequr fp2 move.l (Ptr),d0 ;move.l (a1),d0 fmove.x PI,fp0 ;fmove.x fp2,fp0 The Register List Symbol ------------------------ The register list symbol is defined by `reg' and represents the register mask for `Movem' and `fmovem'. You must not mix FPU and Integer registers with each other in a register list. mask reg d0/d2/d4-d7/a0-a4/a6-a7 mask1 reg d0-a6 mask2 reg d0-6 fmask reg fp0-fp2/fp4-fp5 The Macro Symbol ---------------- By using the command `macro' or `cmacro' after the symbol, the symbol is defined as a macro. The macro block is terminated by the command endm. The Macro cmacro is case-insensitive and therefore useful to emulate commands that are missing from the core. Symbol[:] macro . . Symbol[:] endm Datatypes --------- The assembler understands 3 distinct datatypes. * 32Bit Integer * 96Bit Extended Floating Point * 96Bit Packed Binary Floating Point At the moment only integer datatypes are supported in arithmetic arguments so you can only use the FPU datatypes as constants. +-------------+--------------------+ | Format | Representation | +-------------+--------------------+ +-------------+--------------------+ | Decimal | 1024 | +-------------+--------------------+ | Hexadecimal | $400 | +-------------+--------------------+ | Binary | %10000000000 | +-------------+--------------------+ | Ascii | "OK", 'OK' or `OK` | +-------------+--------------------+ Furthermore you can use symbols or the character `*' that represents the program counter in arguments. There are limitation in the use of symbols in arguments. For example you can only add or subtract constants from an external label, Floatingpoint Values can only be used as simple constants,... By the postfix `k' after decimal value the value is multply by $1000. `Floating Point Format' +-------------+-------------------------------------+ | Format | Representation | +-------------+-------------------------------------+ +-------------+-------------------------------------+ | Extended | '[+,-]3. 145637848298628[e[+,-]123] | +-------------+-------------------------------------+ | Packed | ''[+,-]3. 145637848298628[e[+,-]123]| +-------------+-------------------------------------+ Datatype Conversion ------------------- All commands are performed with these 3 datatypes and then converted into the required datatype. For example a 32Bit integer can be converted into 16Bit and 8 Bit; an extended floating point into a double or single floating point. Floating point datatypes are rounded by a convertation. If a rouding error occurs the parser returns with an error. Datatype Format --------------- Internal Datatype Structure * Integer +--------+---------+ | Bit 31 | 30..0 | +--------+---------+ +--------+---------+ | S | Integer | +--------+---------+ * Single Floating Point +--------+-----------------+------------+ | Bit 31 | Bits 30..23 | Bits 22..0 | +--------+-----------------+------------+ +--------+-----------------+------------+ | Sign | Biased Exponent | Fraction | +--------+-----------------+------------+ * Double Floating Point +--------+-----------------+------------+ | Bit 63 | Bits 62..52 | Bits 51..0 | +--------+-----------------+------------+ +--------+-----------------+------------+ | Sign | Biased Exponent | Fraction | +--------+-----------------+------------+ * Extended Floating Point +--------+-----------------+------------+ | Bit 95 | Bits 94..80 | Bits 62..0 | +--------+-----------------+------------+ +--------+-----------------+------------+ | Sign | Biased Exponent | Mantisse | +--------+-----------------+------------+ * Packed Binary Floating Point +------+------+------+------+------+------+------+------+ | MEYY | EXP2 | EXP1 | EXP0 | EXP3 | 0000 | 0000 | M016 | +------+------+------+------+------+------+------+------+ | M015 | M014 | M013 | M012 | M011 | M010 | M009 | M008 | +------+------+------+------+------+------+------+------+ | M007 | M006 | M005 | M004 | M003 | M002 | M001 | M000 | +------+------+------+------+------+------+------+------+ * M is the sign (+ or -) of the fraction * E is the sign (+ or -) of the exponent * Y are the internal flags for infinity and NAN * E002-000 are the numbers of the exponent from 2 to 0, EXP3 is used internally. * M016-000 are the numbers of the fraction from 16 to 0. Each number lies in the range from 0 to 9.. Operations ---------- Operators --------- +----------+-------------------------------+ | Operator | Function | +----------+-------------------------------+ +----------+-------------------------------+ | `+' | 32Bit signed Addition | +----------+-------------------------------+ | `-' | 32Bit signed Subtraction | +----------+-------------------------------+ | `*' | 32Bit signed Multiplication | +----------+-------------------------------+ | `/' | 32Bit signed Division | +----------+-------------------------------+ | '|' | 32Bit Or | +----------+-------------------------------+ | `!' | 32Bit Or | +----------+-------------------------------+ | `&' | 32Bit And | +----------+-------------------------------+ | `^' | 32Bit Eor | +----------+-------------------------------+ | `<<' | logic 32Bit Shift to the left | +----------+-------------------------------+ | `>>' | logic 32Bit Shift to the right| +----------+-------------------------------+ | `~' | 32Bit Not | +----------+-------------------------------+ Basm cares for the operator priorities but be careful while porting Seka Sources because Seka doesn't care for the priorities. Functions --------- The following functions are supported. * _bitnum(Argument) calculates the bit number of the argument. If this is impossible an error occurs. * _bitfield(Argument) calculates the bit mask of the argument. If this is impossible an error occurs. * _extb(Argument) equal to the 680xx command extb * _extw(Argument) equal to the 680xx command extw * _min(Argument[,Argument,...]) calculate the minimum of the argument. * _max(Argument[,Argument,...]) calculate the maximum of the argument. Assembler Commands ================== Hunk/Link Commands ------------------ [Label] section Name[,Typ[,[RelocModus,Memtyp] .............................................. defines a new logical unit so that the DOS-Loader has the opportuntity to place smaller hunks into free memory blocks. Another use for this command is to set different memory types for the hunk to load gfx data into the chipmem. If you don't specify the section type it assumes "",Code,Public. * Name = Hunkname * Hunk type `CODE' starts a code segment. `DATA' starts a data segment. `BSS' starts an undefined data segment. `DEBUG' starts a custom debug segment. `CUSTOM' starts a Custom-Hunk area. * Reloc-Mode defines the width of the hunk relocation. Default 32Bit `RELOC16' sets the reloc width to 16bit.(V37) `RELOC32' sets the reloc width to 32bit.(Default) * Memtype defines the memory attributes of the hunk. If you add a `_p', `_c', or `_f' upon the type parameter, you cannot use more memtypes.The memtype `DEBUG' does not allow memory attribute suffixes. `PUBLIC' loads the hunk into the memory with the highest priority. Code Suffix _p. `CHIP' loads the hunk into chip memory. Code Suffix _c. `FAST' loads the hunk into fast memory. Code Suffix _f. `ADVISORY' ignores the hunk if the OS doesn't understand the type. A kind of Debug-Hunk that can be used by the OS.(V39) `ATTR=?' loads the hunk into the memory with specified memory attributs. (V37) [Label] code [Name[, Memtyp]] ............................. defines a new code hunk and is equivalent to the command `section ?,code,?'. * Name = Hunkname * Memtype defines the memory attributes for the hunk. `PUBLIC' loads the hunk into the memory with the highest priority. Code Suffix _p. `CHIP' loads the hunk into chip memory. Code Suffix _c. `FAST' loads the hunk into fast memory. Code Suffix _f. `ADVISORY' ignores the hunk if the OS doesn't understand the type. A kind of Debug-Hunk that can be used by the OS.(V39) `ATTR=?' loads the hunk into the memory with specified memory attributs. (V37) [Label] data [Name[, Memtyp]] ............................. defines a new data hunk and is equivalent to the command `section ?,data,?'. * Name = Hunkname * Memtype defines the memory attributes for the hunk. `PUBLIC' loads the hunk into the memory with the highest priority. Code Suffix _p. `CHIP' loads the hunk into chip memory. Code Suffix _c. `FAST' loads the hunk into fast memory. Code Suffix _f. `ADVISORY' ignores the hunk if the OS doesn't understand the type. A kind of Debug-Hunk that can be used by the OS.(V39) `ATTR=?' loads the hunk into the memory with specified memory attributs. (V37) [Label] bss [Name[, Memtyp]] ............................ defines a new BSS hunk and is equivalent to the command `section ?,bss,?'. * Name = Hunkname * Memtype defines the memory attributes for the hunk. `PUBLIC' loads the hunk into the memory with the highest priority. Code Suffix _p. `CHIP' loads the hunk into chip memory. Code Suffix _c. `FAST' loads the hunk into fast memory. Code Suffix _f. `ADVISORY' ignores the hunk if the OS doesn't understand the type. A kind of Debug-Hunk that can be used by the OS.(V39) `ATTR=?' loads the hunk into the memory with specified memory attributs. (V37) [Label] cseg [Name[, Memtyp]] ............................. has the same function as the command `code'. [Label] dseg [Name[, Memtyp]] ............................. has the same function as the command `data'. idnt Name ......... defines the name of the `HUNK_UNIT' hunk in the object file. * Name = Hunkname identify Name ............. defines the name of the actual hunk. * Name = Hunkname BDebugArg Argument .................. defines a parameter in env:BDebugProgram. It doesn't active this function you have to activate by option "-J" in *Note BOPT: MI_BOPT. * Argument = Argument Text smalldata [Register] .................... activates smalldata mode for the hunk. Optionally, you can also define the smalldata register. Default register is `A4'. The program itself must initialize the smalldata register with the address of the smalldata data hunk. bopt w2- ;68020 addressmode warnings off mc68020 ;68020 mode activated smalldata a3 ;Default is A4!!! xref _LinkerDB ;Special linker symbol lea.l _LinkerDB,a3 ;Address of the smalldata data segments move.l #0,(d_test.l,a3) move.l #"TEST",d_test(a3) moveq #0,d0 tst.b array(a3,d0.w) rts section "__MERGED",BSS ;The smalldata data segments are defined ;the following way d_test: ds.l 1 array: ds.b 20 xref Symbol[, Symbol...] ........................ imports a symbol so that you can access symbols that were exported by XDef. The linker resolves these reference during the link process and creates a program file. If the assembler finds a XRef in the source it creates an object file. This decision can be overruled. * Symbol = Name of the importet symbol. xdef Symbol[, Symbol...] ........................ exports a symbol as global so that other object files can import the symbol by XRef. There's no need to define a symbol before you mark them with XDef. If the assembler finds a XRef in the source it creates an object file. This decision can be overruled. * Symbol = Name of the global symbol global Symbol[, Symbol...] .......................... has the same function like XDef. public Symbol[, Symbol...] .......................... has the same function like XDef. output Name ........... sets an output filename. If you don't specify a filename the assembler uses the source filename and adds the appropriate filetype suffix. * Name = Filename objfile ....... has the same function like Output. exeobj ...... writes a program file if you want to overrule the assembler. linkobj ....... writes an object file if you want to overrule the assembler. org Address ........... activates absolute mode. All command that refer to hunk related functions aren't allowed. For example:. section, xdef, xref. The parameter address sets the base address of the created code. * Address = Absolute Address addsym ...... writes a symbol hunk. debug ..... writes a SAS D1 debug hunk to see source level informations while debugging the program through bdebug. Symbol Commands --------------- CArgs [#Offset,]Symbol[,Symbol.w[,Symbol.l] ........................................... defines the symbol offsets for a stack function. The first Symbol starts with the offset 4 but if you like to use a different Offset it's possible to specify one. Then the offset is increased according to the size of the symbol. If the symbol has no size specifier the default size is word. Sorry..i would use a longword here but to be compatible with Devpac i'm forced to use word. cargs Test1.w,Test2.l move.w Test1(a7),d0 ;Test1=4 move.l Test2(a7),d0 ;Test2=4+2=6 Symbol rs[.width] Count ....................... initializes the Symbol with the value of the counter __RS and increases the __RS counter afterwards by Count*Width. You can use this command as a replacement for the include exec/types.i macros to increase the parsing speed. * Width `B' 1 Byte Valuearea: -$80 <= x < $100 `W' 2 Bytes Valuearea: -$8000 <= x < $10000 `L' 4 Bytes Valuearea: -$80000000 <= x < $10000000 `S' 4 Bytes (Single IEEE-Float) `D' 8 Bytes (Double IEEE-Float) `X' 12 Bytes (Extended IEEE-Float) `P' 12 Bytes (Packed BCD-Float) `Q' 16 Bytes (Quadword) Symbol so[.width] Count ....................... This command has the same function like rs with the exception that the Symbol __SO is used instead of the __RS symbol. Internally both symbols are handled equal. Devpac has introduced the symbol __RS and Macro68k knows the functionality by the name __SO. Symbol fo[.width] Count ....................... decreases the counter __FO by Count*Width and initializes the Symbol with the new value. Useful to create the negative local stackframe symbols needed by link. * Width `B' 1 Byte Valuearea: -$80 <= x < $100 `W' 2 Bytes Valuearea: -$8000 <= x < $10000 `L' 4 Bytes Valuearea: -$80000000 <= x < $10000000 `S' 4 Bytes (Single IEEE-Float) `D' 8 Bytes (Double IEEE-Float) `X' 12 Bytes (Extended IEEE-Float) `P' 12 Bytes (Packed BCD-Float) `Q' 16 Bytes (Quadword) rsreset ....... initializes the counter __RS to 0. rsset Value ........... initializes the counter __RS with the Value * Value = New Index clrso ..... has the same function like rsreset clrfo ..... has the same function like foreset. setso Value ........... has the same function like rsset setrs Value ........... has the same function like rsset setfo Value ........... initializes the counter __FO with the Value * Value = New Index Symbol rsval ............ initializes the Symbol with the value of the __RS counter. Symbol soval ............ has the same function like rsval. Symbol foval ............ initializes the Symbol with the value of the __FO counter. Data Commands ------------- align Value ........... aligns the program counter to an address that can be devided by the value. Useful because certain DOS structures have to be aligned on 4 Byte boundaries. For example FileInfoBlock. Furthermore it's also useful to align subroutines on longword boundaries that they fit better into the cache structure. * Value = Align Mask cnop Offset,Align ................. aligns the program counter to an address that can be devided by the Align value and adds the value onto the address. Internally only Align values < 16 are supported. pad.Width Align[,Value] ....................... aligns the program counter to an address that can be devided by the Align*Width and fills the aligned area by the optional mask value. quad .... aligns the program counter to a 16 Byte address. even .... aligns the program counter to an even address. This function is useful if you define an odd sized data area and you need a word aligned for OS data structures or assembler instructions. odd ... aligns the program counter to an odd address. dc[.width] Value[,Value...] ........................... inserts data of the Width into the code. * Width `B' 1 Byte Valuearea: -$80 <= x < $100 `W' 2 Bytes Valuearea: -$8000 <= x < $10000 `L' 4 Bytes Valuearea: -$80000000 <= x < $10000000 `S' 4 Bytes (Single IEEE-Float) `D' 8 Bytes (Double IEEE-Float) `X' 12 Bytes (Extended IEEE-Float) `P' 12 Bytes (Packed BCD-Float) db Value[,Value,...] .................... inserts a byte with a value in the valuearea -$80 <= x < $100. dw Value[,Value,...] .................... inserts a word with a value in the valuearea -$8000 <= x < $10000. dl Value[,Value,...] .................... inserts a longword with a value in the valuearea -$80000000 <= x < $100000000. ub Value[,Value,...] .................... inserts a byte with a value in the valuearea -$80 <= x < $80. uw Value[,Value,...] .................... inserts a word with a value in the valuearea -$8000 <= x < $8000. ul Value[,Value,...] .................... inserts a longword with a value in the valuearea -$80000000 <= x < $80000000. sb Value[,Value,...] .................... inserts a byte with a value in the valuearea -$80 <= x < $100. sw Value[,Value,...] .................... inserts a word with a value in the valuearea -$8000 <= x < $10000. sl Value[,Value,...] .................... inserts a longword with a value in the valuearea -$80000000 <= x < $100000000. pb Value[,Value,...] .................... inserts a byte with a value in the valuearea 0 <= x < $80. pw Value[,Value,...] .................... inserts a word with a value in the valuearea 0 <= x < $8000. pl Value[,Value,...] .................... inserts a longword with a value in the valuearea 0 <= x < $80000000. nb Value[,Value,...] .................... inserts a byte with a value in the valuearea -$80 <= x < 0. nw Value[,Value,...] .................... inserts a word with a value in the valuearea -$8000 <= x < 0. nl Value[,Value,...] .................... inserts a longword with a value in the valuearea -$80000000 <= x < 0. ds[.width] Count[,Value] ........................ defines a memory area with the length Count * Width and fills the area with an optional Value. Default fill value is 0. Is the Count 0 a cnop 0,Width is run. * Width `B' 1 Byte Valuearea: -$80 <= x < $100 `W' 2 Bytes Valuearea: -$8000 <= x < $10000 `L' 4 Bytes Valuearea: -$80000000 <= x < $10000000 `S' 4 Bytes (Single IEEE-Float) `D' 8 Bytes (Double IEEE-Float) `X' 12 Bytes (Extended IEEE-Float) `P' 12 Bytes (Packed BCD-Float) * Count = Length of the memory area. * Value = optional Fill Value. dsb[.width] Count[,Value] ......................... has the same function like ds dsb[.width] Count[,Value] ......................... has the same function like ds blk[.width] Count[,Value...] ............................ has the same function like ds ascii String1[,String2,...] ........................... inserts Strings. cstring String1[,String2,...] ............................. inserts C-Strings. dstring dtype1,dtype2,dtype3 ............................ inserts the current date string. Datentypen * "w" WeekDay * "d" Date * "t" Time dc.b " (" dstring w,d,t dc.b ")" dc.b $a,$d,0 ;=> (Thursday 14-Okt-93 15:32:06) pstring String[,String,...] ........................... inserts a BCPL string. istring String[,String,...] ........................... inserts strings that terminate with a char that has Bit 7 set. bitstream Mask .............. inserts a bitmask for an image object for example. The bits are aligned to bytes. * Mask = Mask is a string that is based only of 0 and 1. bitstream "01001000001" sprintx "Formatstring"[,Value[,...]] .................................... inserts the resulting string into the code. The string isn't terminated by a 0 so that you can add other strings rather easy. * FormatString - is a string in C-Notation so you can use the known C char types n,t,... The following options are allowed. * FormatSyntax - %[flags][width][.limit][length]type * flags - '-' deactivates left side layout. * width - Field Length. If the first char is '0' the field is filled by '0' on the left side. * limit - defines the maximal count of char that can be inserted from a string. Only legal for %s and %b. * length - The size of the datatype. Default is 16-bit for the typs %d,%u and %x. %l is long (32Bit). Attention! The Assembler always pushes a longword on the stack so always use %l if you don't know what you're doing. * type - The following types are supported. b - BSTR, a 32-bit BPTR Pointer on a bytelength string. A NULL BPTR is handled like an empty string. d - signed decimal u - unsigned decimal x - hexadezimal in lower case. X - hexadecimal in upper case. s - String, a 32-bit Pointer on a NULL-terminated Byte-String. A NULL BPTR is handled like an empty string. c - Char * Value - is an argument that has to be resolvable. Listing I/O Commands -------------------- list .... activates the listing output. Has no function if the global listing output wasn't activated. Listing Format: LINE ADDRESS[Flag1] COMMAND-BYTES[Flag2] SOURCE * Flag1 * + shows that the line was created by a macro. * > shows that the Assembler searches the closing ENDC. * < shows that the Assembler searches the closing ENDM. * Flag2 * + shows a line overflow and that Bytes are ignored. Can often happen during data definitions. nolist ...... deactivates the listing output. Has no function if the global listing output wasn't activated. printx "Formatstring"[,Value[,...]] ................................... outputs the string to the current Stdout and works similar as the known C-Printf function. Look at SPRINTF errfile Name ............ defines the filename for the error output. * Name = Filename lisfile Name ............ defines the filename for the listing output. If no error file was defined the error output is also written into the listing file. * Name = Filename Structuring ----------- symbol[:] macro ............... starts a Macro block. endm .... ends a macroblock. mexit ..... ends a macro call. fail .... creates an error. end ... ends the assembling. if Symbol ......... checks if the symbol value is not NULL and assembles the block depending on the success. ifd Symbol .......... checks if the Symbol exists and assembles the block depending on the success. ifnd Symbol ........... checks if the Symbol doesn't exist and assembles the block depending on the success. ifv String .......... This is a privat command that is used for internal functionality and subject to change. Touch an burn! ifnv String ........... This is a privat command that is used for internal functionality and subject to change. Touch an burn! ifmacrod Macro .............. checks if the Macro exists and assembles the block depending on the success. ifmacrond Macro ............... checks if the Macro doesn't exist and assembles the block depending on the success. ifcmacrod CMacro ................ checks if the CMacro exists and assembles the block depending on the success. ifcmacrond CMacro ................. checks if the CMacro doesn't exist and assembles the block depending on the success. ifc Symbol,Symbol ................. compares the first string with the second string and if they are equal the block is assembled. ifnc 'String','String' ...................... compares the first string with the second string and if they differ the block is assembled. if[condition] Symbol=Symbol ........................... compares the first symbol with the second symbol and decides according to the condition if the block is assembled. * Condition = Normal Bcc-Condition Syntax * Symbol = Normal Symbol else .... activates the condition block if the block above wasn't assembled. elseif ...... activates the condition block if the block above wasn't assembled. endc .... defines the end of a condition block. endif ..... defines the end of a condition block. repeat Count ............ repeats the blocks that is located between repeat and endr by the number Count. rept Count .......... has the same function like Repeat procstart ......... defines a function in a Dice-C assembler output and is used to optimize Link and Unlk. This optimize method isn't working yet. procend ....... defines a function in a Dice-C assembler output and is used to optimize Link and Unlk. This optimize method isn't working yet. File I/O Commands ----------------- incdir Dir[,Dir[,...]] ...................... adds directories to the include path list. BASM uses 2 internal path lists and the current directory to find the include and incbin files. First BASM checks for a : character in the filename and if it finds a volume the file is loaded direct instead of searching it through the pathlists. The first path list contains the paths that were defined in the commandline or *Note BOPT: MI_BOPT by the option -i or through incdir. The second path list contains the paths that were defined in global configuration file ENV:BASMOption. The entries of the second list will be removed when the assembler is closed so that the paths are still correct in ARexx-Mode. The first list is removed every pass. * Dir = Name of the Include-Path. Incpath ....... has the same function like incdir. include Name ............ loads the external include file, for example the OS-Includes. If the file is a precompiled include file it's detected automaticlly. Includes are loaded from the editor or cachefile.library. * Name = Filename include2 Name ............. has the same function like include with the exception that the cachefile.library isn't ignored. * Name = Filename incbin Name[,size] .................. inserts the file with the optional length at the current address into the code. Normally used for sounds and graphics. incbin2 Name[,size] ................... has the same function like incbin with the exception that the cachefile.library isn't used. * Name = Name of the data file. ibytes Name[,Length] .................... has the same function like incbin dsbin Name[,Length] ................... defines a memory area with the length of the file specified by the file. Optinally you can defined the maximal file length. * Name = Filename * Length = maximal file length doscmd Name ........... runs the program Name. dc.b 0,"$VER: Fubar 1.0 by Joe User" doscmd "c:date >ram:Temp" incbin ram:Temp doscmd "c:delete ram:Temp" filecom Name ............ sets the file comment of output file. filecom "This is a file comment" pure .... sets the Pure Bit while writing a program file. Miscellaneous ------------- trashreg Reglist ................ defines the registers that are available to the optimizer. * RegList = A normal Registerlist known by Movem. super ..... deactivates Supervisor warnings. mc[Type] ........ defines the processor type to allow certain commands and addressmodes. Processor-Type * 68000 default mode * 68010 * 68020 * 68030 * 68040 * 68060 * 68881 * 68882 bopt [opt[,...],...] .................... sets the assembler options. Options `m1[+,-]' activates/deactivates 68010 mode. `m2[+,-]' activates/deactivates 68020 mode. `m3[+,-]' activates/deactivates 68030 mode. `m4[+,-]' activates/deactivates 68040 mode. `m6[+,-]' activates/deactivates 68060 mode. `mf[+,-]' activates/deactivates 68881/2 mode. `ue[+,-]' activates/deactivates writing an executable file. `uo[+,-]' activates/deactivates writing an object file. `ua[+,-]' activates/deactivates writing an absolut file. `un[+,-]' activates/deactivates writing file. `p[+,-]' activates/deactivates writing a preassembled Include file. `g[+,-]' activates/deactivates adding the prefix _ to each exported symbol. Default is off. `sx[+,-]' activates/deactivates writing all XRef/XDef symbols to a symbol hunk. Default is off. `sl[+,-]' activates/deactivates writing all normal symbols to a symbol hunk. Default is off. `sa[+,-]' activates/deactivates writing all symbols to a symbol hunk. Default is off. `sd[+,-]' activates/deactivates writing a BASM custom format Debug Hunk. Makes only sense as a program file and it needs a lot hd space because it includes all sources. Default is off. `s1[+,-]' activates/deactivates writing a SAS D1 compatible Debug Hunk. Default is off. `sf[+,-]' activates/deactivates writing the full sourcefile path into the debug hunk. You should only use this for your own development system because other users may have different HD layouts. This option has only a meaning with in a SAS D1 Debug Hunk. Default is off. `j[+,-]' activates/deactivates setting the PURE Bit for a program file. The PURE Bit tells the Shell that this program can be loaded resident. Default is off. `J[+,-]' activates/deactivates creating the file ENV:BDebugProgram that contains the assembled filename for BDebug. Default is off. `a[+,-]' activates/deactivates creating of an .info file for each program. Useful if you use the assembler through the WB. Default is off. `A[+,-]' activates/deactivates Arexxmode. Only allowed in the commandline. Default is off. `i<DirName>' defines the include path. `o<FileName>' defines the object filename. If not specified the assembler uses the source file name without the source suffix as the output name. For object files it adds the suffix `.o'. `P<Priority>' sets the task priority. `c[+,-]' activates/deactivates that the assembler interpretes Upper and Lower case as 2 different chars. Default is on. `f[+,-]' activates/deactivates a faster mode that resolves all references in the 2nd pass. Fortunately this mode needs more memory and has some disadvantages like uncorrect values during the listing. This option has no effect during optimizing. Default is off. `F[+,-]' activates/deactivates a faster IO mode that keeps include paths as locks instead of strings. The older basm versions used locks to store paths but as this isn't compatible with multi-assigns I changed that. `M<Bytes>' defines the max macro expansion size. If you get a macromemerror you should increase the size. Default 1000 Bytes. `Z<Address>' tells the assembler that the source is starts in the memory at the defined address. Useful for ARexx scripts. Option is only available in the commandline. `x[+,-]' uses the `cachefile.library' to load resident Includes/Incbins or add unknown files to the cachefile.library database. Default is off. `X[+,-]' erases all files that are controled by the cachefile.library. Default is off. `y[+,-]' shows all files that are controled by the cachefile.library. Default is off. `l[+,-]' activates/deactivates the listing output. Default is off. `l0[+,-]' activates/deactivates the listing macro expansion. Default is on. `L<Listingfile>' defines the Listing filename. `h[+,-]' activates/deactivates the symbol listing output. Default is off. `H[+,-]' activates/deactivates the unused symbol output. Default is off. `v[+,-]' outputs a statistic after assembling. Default is off. `V[+,-]' as little status output as possible Default is off. `e[+,-]' creates an error list. Default is on. `es[+,-]' outputs the error list in the Barfly shell. This option has no meaning in BASM. `wo[+,-]' activates/deactivates Optimizing warnings. Default is on. `ws[+,-]' activates/deactivates Supervisor warnings. Default is on. `wm[+,-]' activates/deactivates Move16 warnings because the use of the move16 command is dangerous if you don't know the problems. Default is on. `wp[+,-]' activates/deactivates pc-relative access to different section warnings. The linker is needed to resolve such files. Default is on. `w2[+,-]' activates/deactivates 68020 addressmode warnings. Default is on. `w4[+,-]' activates/deactivates 64k-Access warnings. It's useful if you accidently avoid to forget the address register. Example: move.l 8,d0 instead of move.l 8(an),d0 Default is on. `b0' sets the Default Branch Length to 8-Bit. .b Default is off. `b1' sets the Default Branch Length to 16-Bit. .w Default is on. `b2' sets the Default Branch Length to 32-Bit. .l Default is off. `B0' sets the Default BaseDisplacement-Width to 8 Bit. .b Default is on. `B1' sets the Default BaseDisplacement-Width to 16 Bit. .w Default is off. `B2' sets the Default BaseDisplacement-Width to 32 Bit. .l Default is off. `n0' sets the Default OuterDisplacement-Width to 16 Bit. .w Default is off. `n1' sets the Default OuterDisplacement-Width to 32 Bit. .l Default is on. `q[+,-]' activates/deactivates align long after each rts, bra or jmp to align blocks to the cache structure. Default is off. `O[+,-]' activates/deactivates the Optimizer. Without this option no optimizing will happen besides addressmode converting. Default is off. `OG[+,-]' activates/deactivates Forward Reference Optimizing to use every possibility. In this mode the source is assembled until no further optimizing method is found. First the source is assembled normally. This is shown by the Output Pass 1. Afterwards the optimize passes are started and continued until no further symbol changes and length errors occur. This can take a while and depends on the source size. Default is off. `OT[+,-]' activates/deactivates Time Optimizing. Default is off. Addressmode Converting `OC0[+,-]' *Note bdwan: OP_DIRECT `OC1[+,-]' *Note bdwpc: OP_DIRECT `OC2[+,-]' *Note anxn: OP_DIRECT `OC3[+,-]' *Note pcxn: OP_DIRECT `OC4[+,-]' *Note bdw: OP_DIRECT `OC5[+,-]' *Note bdl: OP_DIRECT `OC6[+,-]' *Note an: OP_DIRECT `OC7[+,-]' *Note pc: OP_DIRECT `ODD[+,-]' activates Direct Addressmode Optimizing Direct Optimizing `OD0[+,-]' *Note move: OP_OPTIMIZE `OD00[+,-]' *Note move: OP_OPTIMIZE `OD01[+,-]' *Note move: OP_OPTIMIZE `OD02[+,-]' *Note move: OP_OPTIMIZE `OD03[+,-]' *Note move: OP_OPTIMIZE `OD04[+,-]' *Note move: OP_OPTIMIZE `OD05[+,-]' *Note move: OP_OPTIMIZE `OD06[+,-]' *Note move: OP_OPTIMIZE `OD07[+,-]' *Note move: OP_OPTIMIZE `OD08[+,-]' *Note move: OP_OPTIMIZE `OD09[+,-]' *Note move: OP_OPTIMIZE `OD0a[+,-]' *Note move: OP_OPTIMIZE `OD1[+,-]' *Note clr: OP_OPTIMIZE `OD2[+,-]' *Note add: OP_OPTIMIZE `OD3[+,-]' *Note sub: OP_OPTIMIZE `OD4[+,-]' *Note lea: OP_OPTIMIZE `OD5[+,-]' *Note cmp: OP_OPTIMIZE `OD6[+,-]' *Note bcc: OP_OPTIMIZE `OD7[+,-]' *Note jsr: OP_OPTIMIZE `OD8[+,-]' *Note jmp: OP_OPTIMIZE `OD9[+,-]' *Note asl: OP_OPTIMIZE `ODa[+,-]' *Note or: OP_OPTIMIZE ( This Optimizing is deactivated internal ) `ODb[+,-]' *Note eor: OP_OPTIMIZE ( This Optimizing is deactivated internal ) `ODc[+,-]' *Note and: OP_OPTIMIZE `ODd[+,-]' *Note mulu: OP_OPTIMIZE `ODe[+,-]' *Note muls: OP_OPTIMIZE `ODf[+,-]' *Note jsr+rts: OP_OPTIMIZE `ODg[+,-]' *Note jmp+rts: OP_OPTIMIZE `ODh[+,-]' *Note MovemNoRegister: OP_OPTIMIZE `ODi[+,-]' *Note MovemOneRegister: OP_OPTIMIZE `ODj[+,-]' *Note Link: OP_OPTIMIZE `OAP[+,-]' activates *Note PC-Relative: OP_ADDRESS Optimizing `OAS[+,-]' activates *Note Smalldata: OP_ADDRESS Optimizing `OAL[+,-]' activates *Note long to word: OP_ADDRESS Optimizing `OAX[+,-]' activates *Note x(An) to (An): OP_ADDRESS Optimizing `OAY[+,-]' activates *Note 68020 An-EA: OP_ADDRESS Optimizing `OAZ[+,-]' activates *Note 68020 PC-EA: OP_ADDRESS Optimizing `OAR[+,-]' activates *Note Register: OP_REGISTER Optimizing You should be careful with the command *Note BOPT: MI_BOPT when you activate Global-Optimize. In every parse the default config is set and therefore you should define all global options in the commandline or in the configuration file. 680xx Meta Commands ------------------- mb Operand1,Operand2 .................... has the same function as move.b. mw Operand1,Operand2 .................... has the same function as move.w. ml Operand1,Operand2 .................... has the same function as move.l. mq Operand1,Operand2 .................... has the same function as moveq. xor.? Operand1,Operand2 ....................... has the same function as eor.?. xori.? Operand1,Operand2 ........................ has the same function as eori.?. bhs.? Label ........... has the same function as bcc.?. blo.? Label ........... has the same function as bcs.?. Assembler Macros ================ Macros are meta commands that can be based of many assembler instructions to achieve an abstracter source layout. In a macro you can use several different pattern that are replaced by appropriate parameters when the macro is called. The parameter that are passed during a macro call are represented by the following patterns: \0,...,\9, \a,...,\z, \A,...,\Z. The pattern ids are using the hexadecimal format. If a pattern is used with no related parameter an empty string is inserted. Furthermore if a parameter contains tabulators or spaces it has to be placed between <...>. When a macro needs relative labels and is should be called more than one time you should use the special pattern @. This pattern is replaced by a number that is based of 4 digits and that is increased after each call. The pattern \# is replaced by the value of the symbol narg that represents the count of macro parameters. Besides the standard patterns there are some more advanced pattern functions supported that look like \*Function-Name. These functions don't belong to the motorola standard thus they aren't supported by every assembler. Another important point is that you can also call macros from from macros but you can't define macros in macros. The standard macro pattern Label & [. string] & [, string] & [, string] & [,...] & [\\0] & [\\1|] & [\\2|] & [\\3|]...[\\n|] The advanced macro pattern functions * ` \(Argument)' inserts the string of the macroparameter with the number the argument defines. \(1) = \1 \(1+3+4) = \8 * `\*upper(String)' inserts the string in upper case. * `\*lower(String)' inserts the string in lower case. * `\*valof(Argument)' inserts the decimal value of the argument as a string. * `\*strlen(Symbol)' inserts the length of a symbol as a string. * `\*right(String,n)' inserts n chars of the right side of the string. If the string contains less than n chars the whole string is inserted. * `\*left(String,n)' inserts n chars of the left side of the string. If the string contains less than n chars the whole string is inserted. * `\*mid(String,n,m)' inserts chars from position n to m from the string. If the position is outside of the string length the chars till the end of the string is inserted. openwind MACRO move.l intbase,a6 lea.l \1,a0 jsr OpenWindow(a6) ENDM start: openwind newwindow movewind MACRO move.l intbase,a6 move.l \1,a0 moveq #0,d0 move.\0 \2,d1 IFC `\0',`b' ext.w d1 ENDC jsr MoveWindow(a6) ENDM start: move.b #10,d2 1$: movewind.b newwindow,d2 addq.b #1,d2 cmp.b #100,d2 bne.s 1$ wait MACRO moveq #-1,d0 wait\ dbra d0,wait\ ENDM start: wait wait wait test MACRO move.l #\*upper(Hello),d0 move.l #\*lower(Hello),d0 move.l #\*strlen(1234567890123456),d0 move.l #\*valof(value),d0 rts cstring "\*left(abcdefgh,4)" even cstring "\*left(abcdefgh,10)" even cstring "\*right(abcdefgh,4)" even cstring "\*right(abcdefgh,10)" even cstring "\*mid(abcdefgh,2,4)" even cstring "\*mid(abcdefgh,2,8)" even ENDM value = 123456789 hello: test value = 123456789 hello: move.l #HELLO,d0 move.l #hello,d0 move.l #16,d0 move.l #123456789,d0 rts cstring "abcd" even cstring "abcdefgh" even cstring "efgh" even cstring "abcdefgh" even cstring "cdef" even cstring "cdefgh" even PUTTAG MACRO IFC "\2","" PUTTAG_COUNT set 0 ENDC IFNC "\2","" move.l \2,-(a7) PUTTAG_COUNT SET PUTTAG_COUNT+4 ENDC move.l \1,-(a7) PUTTAG_COUNT SET PUTTAG_COUNT+4 IFC "\1","#TAG_END" PUTTAG_COUNT SET 4 ENDC ENDM CLEARTAG MACRO lea.l PUTTAG_COUNT(a7),a7 ENDM PUTTAG #TAG_END PUTTAG #WA_Width,#100 PUTTAG #WA_ScreenTitle,#Title . . move.l a7,a1 sub.l a0,a0 jsr OpenWindowTagList(a6) CLEARTAG Highlevel Macros ================ In highlevel macros the operands are based of legal addressmodes. Arguments are based of operands and the operators +,-,<<,>>. Conditions are based of !,=, <,>,<=,>=,<>. By using highlevel macros you can make the programming of non critical source areas easier and more abstract. Blame Mike Schwartz for this idea...he forced me to do it:-B .REG .... sets the accumulator register that is used to calculate arguments. Default register is D0. .BRANCH b|w|l ............. sets the length of branch commands that are used in the highlevel macros. Standard length is .b. .FOR Operand[.b|w|l] = Operand TO Operand STEP Operand ...................................................... creates code for a for loop. The optional width you define after the first operand sets the width for all operations in the for loop. .FOR d0.w = #1 to STEP #2 addq.w #1,d1 .NEXT ;Compiled Code move.w #1,d0 __for1: addq.w #1,d1 add.w #2,d0 cmp.w ,d0 blt.b __for1 .NEXT ..... closes the outer .FOR loop. .IF [Argument] =,!, < , > , <> Operand ...................................... creates code for an IF-Operation. You can remove the first argument if you want to test the operand. For example .IF <> .IF (a0) + #0 <> d1 moveq #0,d0 .ELSE moveq #1,d0 .ENDIF ;Compiled Code move.l (a0),d7 add.l #0,d7 cmp.l d1,d7 beq.b __else1 moveq #0,d0 bra.b __endif1 __else1: moveq #1,d0 __endif1: .ELSE ..... starts an alternative IF-Block. .ENDIF ...... closes the outer .IF block. .WHILE [Argument] =,!, < , > , <> Operand ......................................... creates code for a while loop The optional width you define after the first operand sets the width for all operations in the while loop. .WHILE d0 <> #0 addq.w #1,d1 .ENDWHILE ;Compiled Code __while1: cmp.l #0,d0 beq.s __endwhile1 addq.w #1,d1 bra.s __while1 __endwhile1: .ENDWHILE ......... closes the outer while loop. .CALL Function [, Argument [, Argument [,...]]] ............................................... calls a C-Function by parsing the arguments through the stack. Arguments are calculated in the accumulator register. .CALL func , test + 0 - #20 , #test ;Compiled Code move.l test,d7 add.l 0,d7 sub.l #20,d7 move.l d7,-(a7) move.l #test,-(a7) jsr func ifnc "8","0" ;Were there any parameters ? lea.l __CALLSize(a7),a7 endc .RETURN Argument ................ returns a result value in the accumulator register. .return d1 + d2 + #$100 ;Compiled Code move.l d1,d7 add.l d2,d7 add.l #$100,d7 .DEF func [, Operand [, Operand [,...]]] ........................................ defines a C-Stack function and loads the defined parameters into the operands. .DEF func , d0.w , d1 , (a0) .ENDDEF ;Compiled Code XDEF func link a5,#0 move.w $0a(a5),d0 move.l $0c(a5),d1 move.l $10(a5),(a0) unlk a5 ;.ENDDEF rts .ENDDEF ....... closes a function that was started by .DEF .LET [ Operand =] Argument .......................... calculates an argument in an accumulator or moves the value to a defined operand. .LET + 4 - #LN_SIZE << #7 .LET d1 = (a1) - (a0) ;Compiled Code add.l 4,d7 sub.l #LN_SIZE,d7 lsl.l #7,d7 move.l (a1),d7 sub.l (a0),d7 move.l d7,d1 Predefined Symbols ================== NARG .... represents the macro parameter count in a macro. BARFLY ...... represents the assembler version. 680xx ..... represents the CPU processor type. 6888x ..... represents the FPU processor type. _MOVEMBYTES ........... represents the byte count the last movem transfer used. lea _MOVEMBYTES(a7),a7 ;frees the stack _MOVEMREGS .......... represents the last movem register mask. movem (a7)+,_MOVEMREGS __RS .... represents the RS-Counter. __SO .... represents the RS-Counter. __FO .... represents the FO-Counter. Optimizing ========== Optimize Methods... Direct Addressmode Optimizing ----------------------------- The assembler can direct optimize certain 68020...60 Addressmodes if a faster 68000 addressmode exists. This optimizing method should always be activated because of compatibility reasons. * `(bd.w,an)' can be optimized to `x(an)' that removes 1 word and some cycles. +------------------------+--------------------+--------+ | Addressmode | Optimizing | Option | +------------------------+--------------------+--------+ +------------------------+--------------------+--------+ | move. l (1000.w,an),dn | move.l 1000(an),dn | -OC0 | +------------------------+--------------------+--------+ * `(bd.w,pc)' can be optimized to `x(pc)' that removes 1 word and some cycles. +------------------------+--------------------+--------+ | Addressmode | Optimizing | Option | +------------------------+--------------------+--------+ +------------------------+--------------------+--------+ | move. l (1000.w,pc),dn | move.l 1000(pc),dn | -OC1 | +------------------------+--------------------+--------+ * `(bd.w)' can be optimized to `bd.w' that removes 1 word and some cycles. +------------------------+--------------------+--------+ | Addressmode | Optimizing | Option | +------------------------+--------------------+--------+ +------------------------+--------------------+--------+ | move. l (bd.w),dn | move.l bd.w,dn | -OC4 | +------------------------+--------------------+--------+ * `(bd.l)' can be optimized to `bd.l' that removes 1 word and some cycles. +------------------------+--------------------+--------+ | Addressmode | Optimizing | Option | +------------------------+--------------------+--------+ +------------------------+--------------------+--------+ | move. l (bd.l),dn | move.l bd.l,dn | -OC5 | +------------------------+--------------------+--------+ * `(an)' can be optimized to `(an)' that removes 1 word and some cycles. The addressmode `(an)' can be interpreted as a subgroup of `(bd,an,xn)'. Because `(an)' is a normal 68000 addressmode you should never switch off this optimizing method. +------------------------+--------------------+--------+ | Addressmode | Optimizing | Option | +------------------------+--------------------+--------+ +------------------------+--------------------+--------+ | move. l (an),dn | move.l (an),dn | -OC6 | +------------------------+--------------------+--------+ * `(pc)' can be optimized to `(pc)' that removes 1 word and some cycles. The addressmode `(pc)' can be interpreted as a subgroup of `(bd,pc,xn)'. Because `(pc)' is a normal 68000 addressmode you should never switch off this optimizing method. +------------------------+--------------------+--------+ | Addressmode | Optimizing | Option | +------------------------+--------------------+--------+ +------------------------+--------------------+--------+ | move. l (pc),dn | move.l (pc),dn | -OC7 | +------------------------+--------------------+--------+ Address Optimizing ------------------ * `Long' +---------------+----------------+---------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+---------------------+--------+ +---------------+----------------+---------------------+--------+ | x.l,EA | x.w,EA | $ffff8000<=x<=$7fff | | +---------------+----------------+---------------------+ -OAL | | EA,x.l | EA,x.l | $ffff8000<=x<=$7fff | | +---------------+----------------+---------------------+--------+ * `x(an)' +---------------+----------------+---------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+---------------------+--------+ +---------------+----------------+---------------------+--------+ | x(an),EA | (an),EA | x=0 | | +---------------+----------------+---------------------+ -OAX | | EA,x(an) | EA,(an) | x=0 | | +---------------+----------------+---------------------+--------+ * `PC-Relative' +---------------+----------------+---------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+---------------------+--------+ +---------------+----------------+---------------------+--------+ | label,EA | label(pc),EA | $ffff8000<=x<=$7fff | -OAP | +---------------+----------------+---------------------+--------+ * `A4-Smalldata' +---------------+----------------+---------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+---------------------+--------+ +---------------+----------------+---------------------+--------+ | x.l,EA | x(a4),EA | $ffff8000<=x<=$7fff | | +---------------+----------------+---------------------+ -OAS | | EA,x.l | EA,x(a4) | $ffff8000<=x<=$7fff | | +---------------+----------------+---------------------+--------+ * `68020-An' +---------------+----------------+---------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+---------------------+--------+ +---------------+----------------+---------------------+--------+ | (x.l,an) | (an) | x=0 | -OAY | +---------------+----------------+---------------------+ | | (x.l,an) | x(an) | $ffffff80<=x<=$7f | | +---------------+----------------+---------------------+ | | (x,an,xn) | 0(an,xn) | x=0 | | +---------------+----------------+---------------------+ | | (x,an,xn) | (x.b,an,xn) | $ffffff80<=x<=$7f | | +---------------+----------------+---------------------+ | | (x,an,xn) | (x.w,an,xn) | $ffff8000<=x<=$7fff | | +---------------+----------------+---------------------+ | | ([?],x) | ([?]) | x=0 | | | | | ? is also optimized | | +---------------+----------------+---------------------+ | | ([?],x) | ([?],x.w) | $ffff8000<=x<=$7fff | | | | | ? is also optimized | | +---------------+----------------+---------------------+ | | ([?],xn,x) | ([?],xn) | x=0 | | | | | ? is also optimized | | +---------------+----------------+---------------------+ | | ([?],xn,x) | ([?],xn,x.w)| $ffff8000<=x<=$7fff | | | | | ? is also optimized | | +---------------+----------------+---------------------+--------+ * `68020-PC' +---------------+----------------+---------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+---------------------+--------+ +---------------+----------------+---------------------+--------+ | (x.l,pc) | (pc) | x=0 | -OAZ | +---------------+----------------+---------------------+ | | (x.l,pc) | x(pc) | $ffffff80<=x<=$7f | | +---------------+----------------+---------------------+ | | (x,pc,xn) | 0(pc,xn) | x=0 | | +---------------+----------------+---------------------+ | | (x,pc,xn) | (x.b,pc,xn) | $ffffff80<=x<=$7f | | +---------------+----------------+---------------------+ | | (x,pc,xn) | (x.w,pc,xn) | $ffff8000<=x<=$7fff | | +---------------+----------------+---------------------+ | | ([?],x) | ([?]) | x=0 | | | | | ? is also optimized | | +---------------+----------------+---------------------+ | | ([?],x) | ([?],x.w) | $ffff8000<=x<=$7fff | | | | | ? is also optimized | | +---------------+----------------+---------------------+ | | ([?],xn,x) | ([?],xn) | x=0 | | | | | ? is also optimized | | +---------------+----------------+---------------------+ | | ([?],xn,x) | ([?],xn,x.w)| $ffff8000<=x<=$7fff | | | | | ? is also optimized | | +---------------+----------------+---------------------+--------+ #x Optimizing ------------- * `Move' +-------------------+------------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +-------------------+------------------+-------------------------+--------+ | | | | -OD0 | +-------------------+------------------+-------------------------+--------+ | move.l #x,dn | moveq #x,dn | $ffffff80<=$7f | -OD00 | +-------------------+------------------+-------------------------+ | | move.? #0,an | suba.l an,an | ? = w or l | -OD01 | +-------------------+------------------+-------------------------+ | | move.l #x,dn | moveq #y,dn | $10000<=x<=$7f0000 | -OD02 | | | swap dn | | | +-------------------+------------------+-------------------------+ | | move.l #x,dn | moveq #y,dn | $ff80ffff<=x<=$fffEffff | -OD03 | | | swap dn | | | +-------------------+------------------+-------------------------+ | | move.l #x,dn | moveq #y,dn | $80<=x<=$ff | -OD04 | | | not.b dn | | | +-------------------+------------------+-------------------------+ | | move.l #x,dn | moveq #y,dn | $ffff<=x<=$ff81 | -OD05 | | | neg.w dn | | | +-------------------+------------------+-------------------------+ | | move.l #x,dn | moveq #y,dn | $ffff0080<=x<=$ffff0001 | -OD06 | | | neg.w dn | | | +-------------------+------------------+-------------------------+ | | move.? #0,EA | clr.? EA | ? = w or l.See Trashreg | -OD07 | | | | optimizing. I also check| | | | | if it accesses the HW | | +-------------------+------------------+-------------------------+ | | move.b #$ff,EA | st EA | | -OD08 | +-------------------+------------------+-------------------------+ | | movea.l -4(an),an | movea.l -(an),an | | -OD09 | +-------------------+------------------+-------------------------+ | | movea.w -2(an),an | movea.w -(an),an | | -OD0a | +-------------------+------------------+-------------------------+--------+ * `Clr' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | clr.l dn | moveq #0,dn | | -OD1 | +---------------+----------------+-------------------------+--------+ * `Add' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | add.? #x,EA | addq.? #x,EA | 1<=x<=8 | -OD2 | +---------------+----------------+-------------------------+ | | add.? #x,EA | subq.? #x,EA | -8<=x<=-1 | | +---------------+----------------+-------------------------+ | | add.? #x,an | lea.l x(an),an | $ffff8000<=x<=$7fff | | +---------------+----------------+-------------------------+ | | add.? #0,EA | tst.? EA | legal EA | | +---------------+----------------+-------------------------+ | | add.? #0,an | removed | | | +---------------+----------------+-------------------------+--------+ * `Sub' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | sub.? #x,EA | subq.? #x,EA | 1<=x<=8 | -OD3 | +---------------+----------------+-------------------------+ | | sub.? #x,EA | addq.? #x,EA | -8<=x<=-1 | | +---------------+----------------+-------------------------+ | | sub.? #x,an |lea.l -x(an),an | $ffff8000<=x<=$7fff | | +---------------+----------------+-------------------------+ | | sub.? #0,EA | tst.? EA | legal EA | | +---------------+----------------+-------------------------+ | | sub.? #0,an | removed | | | +---------------+----------------+-------------------------+--------+ * `Lea' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | lea x(an),an | addq.w #x,an | 1<=x<=8 | | +---------------+----------------+-------------------------+ -OD4 | | lea x(an),an | subq.w #x,an | -8<=x<=-1 | | +---------------+----------------+-------------------------+--------+ * `Cmp' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | cmp.? #0,EA | tst.? EA | | -OD5 | +---------------+----------------+-------------------------+--------+ * `Bcc' The assembler tries to optimize the branch on the smallest possible length so that can win max 2 words and some cycles. +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | Bcc.l label | Bcc.w label | $8000<=label<=$7fff | -OD6 | +---------------+----------------+-------------------------+ | | Bcc.l label | Bcc.s label | $80<=label<=$7f | | +---------------+----------------+-------------------------+ | | Bcc.w label | Bcc.s label | $80<=label<=$7f | | +---------------+----------------+-------------------------+--------+ Attention! This optimizing methid is unsafe when you use BRANCH-Tables. You should switch off the optimize method over this area. * `Jsr' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | jsr label | bsr.w label | $8000<=Offset<=$7fff | | +---------------+----------------+-------------------------+ -OD7 | | jsr label | bsr.s label | $80<=Offset<=$7f | | +---------------+----------------+-------------------------+--------+ Attention! This optimizing methid is unsafe when you use JSR-Tables. You should switch off the optimize method over this area. * `Jmp' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | jmp label | bra.w label |$ffff8000<=Offset<=$7fff | | +---------------+----------------+-------------------------+ -OD8 | | jmp label | bra.s label |$ffffff80<=Offset<=$7f | | +---------------+----------------+-------------------------+--------+ * `Asl' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | asl.? #1,dn | add.? dn,dn | | -OD9 | +---------------+----------------+-------------------------+--------+ * `Or' This optimizing method isn't safe because of the changed condition flags. +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | or.? #x,dn | bset #y,dn | x=y^2 | -ODa | +---------------+----------------+-------------------------+--------+ * `Eor' This optimizing method isn't safe because of the changed condition flags. +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | eor.? #x,dn | bchg #y,dn | x=y^2 | -ODb | +---------------+----------------+-------------------------+--------+ * `Mulu' Be very careful with this optimizing. +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | mulu.w #x,dn | swap dn | x=2^y | -ODc | | | clr.w dn | y=y1+y2 | | | | swap dn | y=1,add.l dn,dn | | | | lsl.l #y1,dn | | | | | lsl.l #y2,dn | | | +---------------+----------------+-------------------------+ | | mulu.l #x,dn | lsl.l #y1,dn | x=2^y | | | | lsl.l #y2,dn | y=y1+y2 | | | | | y >= 16 | | | | | swap dn , y-16 | | +---------------+----------------+-------------------------+--------+ * `Muls' Be very careful with this optimizing. +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | muls.w #x,dn | ext.l dn | x=2^y | -ODd | | | asl.l #y1,dn | y=y1+y2 | | | | asl.l #y2,dn | y=1 , add.l dn,dn | | +---------------+----------------+-------------------------+ | | muls.l #x,dn | asl.l #y1,dn | x=2^y | | | | asl.l #y2,dn | y=y1+y2 | | | | | y >= 16 | | | | | swap dn ,y-16 | | +---------------+----------------+-------------------------+--------+ * `Jsr+Rts' +--------------+---------------+------------------------+-------+ | Addressmode | Optimizing | Note | Option | +--------------+---------------+------------------------+-------+ +--------------+---------------+------------------------+-------+ | jsr EA | jmp EA | No optimizing if there's| -ODe | | rts | | a label before RTS | | +--------------+---------------+------------------------+-------+ * `Bsr+Rts' +--------------+---------------+------------------------+-------+ | Addressmode | Optimizing | Note | Option | +--------------+---------------+------------------------+-------+ +--------------+---------------+------------------------+-------+ | jmp EA | jmp EA | No optimizing if there's| -ODf | | rts | | a label before RTS | | +--------------+---------------+------------------------+-------+ * `MovemNoRegister' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | movem.l ,EA | Removed | | | +---------------+----------------+-------------------------+ -ODh | | movem.l EA, | Removed | | | +---------------+----------------+-------------------------+--------+ * `MovemOneRegister' +---------------+----------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+----------------+-------------------------+--------+ +---------------+----------------+-------------------------+--------+ | movem.l Xn,EA | Move.l Xn,EA | Alter the status flags! | | +---------------+----------------+-------------------------+ -ODi | | movem.l EA,Xn | Move.l EA,Xn | Alter the status flags! | | +---------------+----------------+-------------------------+--------+ Register Optimizing ------------------- * `#xxx' is switched off. * An address register is set free by `trashreg'. +---------------+--------------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+--------------------+-------------------------+--------+ +---------------+--------------------+-------------------------+--------+ |move.? EA,label| lea.l label(pc),an | $ffff8000<=label<=$7fff | | | | move.? EA,(an) | | -OAR | +---------------+--------------------+-------------------------+ | | tst.? label | lea.l label(pc),an | $ffff8000<=label<=$7fff | | | | tst.? (an) | | | +---------------+--------------------+-------------------------+ | | not.? label | lea.l label(pc),an | $ffff8000<=label<=$7fff | | | | not.? (an) | | | +---------------+--------------------+-------------------------+ | | neg.? label | lea.l label(pc),an | $ffff8000<=label<=$7fff | | | | neg.? (an) | | | +---------------+--------------------+-------------------------+ | | negx.? label | lea.l label(pc),an | $ffff8000<=label<=$7fff | | | | negx.? (an) | | | +---------------+--------------------+-------------------------+ | | nbcd label | lea.l label(pc),an | $ffff8000<=label<=$7fff | | | | nbcd (an) | | | +---------------+--------------------+-------------------------+ | | scc label | lea.l label(pc),an | $ffff8000<=label<=$7fff | | | | scc (an) | | | +---------------+--------------------+-------------------------+--------+ * `#x' Optimizing on. * A data register is set free by `trashreg'. +---------------+--------------------+-------------------------+--------+ | Addressmode | Optimizing | Note | Option | +---------------+--------------------+-------------------------+--------+ +---------------+--------------------+-------------------------+--------+ | move.l #x,EA | moveq #x,dn | $ffffff80<=x<=$7f | | | | move.l dn,EA | | -OAR | +---------------+--------------------+-------------------------+ | | ori.l #x,EA | moveq #x,dn | $ffffff80<=x<=$7f | | | | or.l dn,EA | | | +---------------+--------------------+-------------------------+ | | eori.l #x,EA | moveq #x,dn | $ffffff80<=x<=$7f | | | | eor.l dn,EA | | | +---------------+--------------------+-------------------------+ | | andi.l #x,EA | moveq #x,dn | $ffffff80<=x<=$7f | | | | and.l dn,EA | | | +---------------+--------------------+-------------------------+ | | addi.l #x,EA | moveq #x,dn | $ffffff80<=x<=$7f | | | | add.l dn,EA | | | +---------------+--------------------+-------------------------+ | | subi.l #x,EA | moveq #x,dn | $ffffff80<=x<=$7f | | | | sub.l dn,EA | | | +---------------+--------------------+-------------------------+ | | cmpi.l #x,EA | moveq #x,dn | $ffffff80<=x<=$7f | | | | cmp.l EA,dn | | | +---------------+--------------------+-------------------------+ | | move.? #0,EA | moveq #0,dn | Time optimizing | | | | move.l dn,EA | must be on | | +---------------+--------------------+-------------------------+--------+ How does Optimizing work ? -------------------------- In single-pass Optimizing the assembler can only optimize commands where it can resolve the reference in the first pass. This means the label or symbol has to be known before. In multi-pass Optimizing it can optimize every command without bothering where the label is defined. The Assembler keeps all labels each pass but increases a change counter if the old contents differs with the new contents. The exception is that the assembler can't optimize commands that depend on symbols that are defined after the command. The reason is that the Assembler has to remove each pass every symbol to avoid problems with `IFD' and `IFND' that can cause that certain areas aren't assembled in multi-pass mode. You probably noticed that i assume that nobody used `IFD' or `IFND' with labels because that would also break multi-pass. An example for a construct that can't be optimized. move.l #1,NULL(a0) NULL=0 Problems... ----------- You should always be careful with optimizing because it can cause bugs in certain source areas. Branch optimizing for example has to be switched off if you use JMP-Towers. lsl.w #1,d0 jmp Tower(pc,d0.w) Tower: bra.w func1 bra.w func2 bra.w func3 could be be optimized to Tower: bra.w func1 bra.s func2 bra.s func3 that leads to program bugs. Solution: bopt OD6- Tower: bra.w func1 bra.w func2 bra.w func3 bopt OD6+ Preassembled Includes ===================== If you want to assemble a program that needs to load a lot includes it's useful to preassemble the includes and load one file because the real slowdown factor is the need loading time. You can only use absolut symbols and macros in a preassembled file. All relative and symbols defined by `set' aren't written into a preassembled file. The created file isn't compress to avoid any slowdown but if the file size is critical you can compress the file by xpk and load it through the xfh filesystem. basm -p Source.S creates the preassembled file Source.p An error location could be absolut symbols that are calculated by relative symbols. You should avoid these symbols. Symbol=Label1-Label2 Resident includes ================= `BASM' can control an Include and Incbin database by the library `cachefile.library' to get rid of the loading delays. The files in the database can be shown and deleted. Basm Assembler System ===================== Cli Calling Convention ---------------------- `Format:' BASM [-Option] Name This is the commandline version of the assembler and can be easy integrated in own development system, for example `Make' and `CED'. An assembler error is indicated by the result 20 and the result 10 is used if no source file were specified. * `Option' The same options are accepted that are described in the assembler command *Note BOPT: MI_BOPT. The following options are accepted additionally. * A[+,-] Turns ARexx mode on/off * C <Configuration> loads a configuration file * d <Symbol=Value> defines a symbol * `Standard-Optionen' ;All other options are deactivated. c+,e+,m1000,wo+,ws+,wm+,w2+,w4+,wp+ b1+,B0+,n1+ OC0+,OC1+,OC2+,OC3+,OC4+,OC5+,OC6+,OC7+, ODD+,OD0+,OD1+,OD2+,OD3+,OD4+,OD5+,OD6+,OD7+,OD8+, OD9+,ODc+,ODe+,ODf+,ODg+ OAP+,OAL+,OAX+,OAY+,OAZ+,OAR+ Configuration ------------- Global ...... You can define the global configuration through the file `ENV:BASMOption'. The internal standard configuration is not replaced but can only be changed. File `ENV:BASMOption' -v -f -c- -iASM: If a line doesn't start with - it's ignored. WB Tooltypes ............ Additionally you can also define the above described configuration options in the tooltypes of the source file icon. Furthermore `BASM' allows a special tooltype to define an output window. o Window= <Window Defintion> ARexx ===== The `BASM' ARexx Port Name is `rexx_BASM' and the ARexx Script suffix `.basm'. To activate the `BASM' ARexx mode you have to start `BASM' with the option -A. BASM .... `BASM [-Option] Name' This ARexx command starts the assembler and coincides with the CLI-syntax structure. BEND .... `BEND' This ARexx command closes the ARexx port and shuts down the assembler. BGETERROR ......... `BGETERROR' With this ARexx command you will receive an explanation of the actual errors. If no errors exist it will return a status code 20. Error Format String OFFSET|FILE|<Error Description> BNEXTERROR .......... `BNEXTERROR' This ARexx command will cause a jump to the next error in the list. If there are no further errors in the list it will return a status code 20. BINITERROR .......... `BINITERROR' This ARexx command will cause a jump to the first entry in the error list. If no error exists it will return a status code 20. BGETWARNING ........... `BGETWARNING' works like `BGETERROR' only for Warnings. BNEXTWARNING ............ `BNEXTWARNING' works like `BNEXTERROR' only for Warnings. BINITWARNING ............ `BINITWARNING' works like `BINITERROR' only for Warnings. Compatibility ============= to other assemblers... ---------------------- Fortunately `BASM' can't be 100% compatible to every assembler on the amiga market. Thus you can expect problems with different sources. In general you can expect problems with commands that don't belong to a standard like option commands. Furthermore you should also be careful with sources that directly depend on the assembler implementation. Because `BASM' is a 1-Pass Assembler in the normal mode with an additional backpatch phase you shouldn't define symbols that can't be resolved at once. An ideal example for this practice is the `Xoper2.2' Source that was developed with the PD Assembler A68k. While assembling with `BASM' the assembler detects that a not defined symbols is accessed through the `SET' command. Generally this should cause an error at once but unfortunately `A68k' doesn't show anything and uses the last value of `cmdnum'. ADDCMD MACRO cmdnum set cmdnum+1 dc.b \1,0 ENDM . . . ;Here it's using `cmdnum' although ;the symbol wasn't defined yet addq #1,d2 cmp.w #cmdnum,d2 bne.s 1$ . . . ;Here the cmdnum is first defined cmdnum set 0 commds ADDCMD 'time' ADDCMD 'taskpri' ADDCMD 'info' ADDCMD 'pri' ADDCMD 'flush' Because the A68k is a 2-Pass Assembler he can assemble this Source without problems. Another Problem is that the assembler argument parser doesn't detect Overflow because of speed reasons. I don't think it's worth it...if you differ tell me your opinion. The assembler doesn't support the following motorola syntax bugs because of the internal structure of the parser it would cause major problems in the multi-pass mode. symbol: equ 0 symbol: equr d0 C-Compiler Assembler -------------------- Dice .... If you use Basm as a `DASM' replacement you have to run `Basm' with the option -OAS to activate the Smalldata mode. If you want to emulate the advanced `Link', `UnLink', `Movem' optimizing `DASM' supports you have to use the options -O, -OG, -ODh, -ODi and -ODj. The option -OG is needed because the link stackframe register list symbols are defined after the commands so the assembler doesn't know them in 1 pass mode. Sorry...i had to disable this mode because i later detected that i have to keep track of the used registers. I'll try to fix this in a later version Literature ********** * [Addison Wesley] RKM Libraries 2.04,CATS * [Addison Wesley] RKM Devices 2.04,CATS * [Addison Wesley] RKM Autodocs\&Includes 2.04,CATS * [Addison Wesley] RKM Hardware 2.04,CATS * [Addison Wesley] RKM Styleguide,CATS * [Addison Wesley] RKM Libraries 1.1,CATS * [Addison Wesley] RKM Intuition 1.1,CATS * [Addison Wesley] RKM Exec 1.1,CATS * [Addison Wesley] RKM Hardware 1.1,CATS * [Edotronik] Kommentiertes Rom-Listing 1,Dr. Ruprecht * [Edotronik] Kommentiertes Rom-Listing 2,Dr. Ruprecht * [Edotronik] Kommentiertes Rom-Listing 3,Dr. Ruprecht * [Ralph Babel] Guru Book,Selbstvertrieb Software ******** For the development of Barfly the following programs were used: * [CATS] Developer CD V2.0 * [B.Hawes] WShell V2.0 * [M.Sinz] Enforcer * [R.Schmidt] CyberGuard * [C.Scheppner] Mungwall * [SAS Institute] SAS/C * [GNU] GCC * [ASDG] CED * [Georg Hessmann] PasTex * [Stefan Stuntz] MFR 2.0d * [Mathias Scheler] Filer * [Matthew Dillon] DNet * [Brian Cerveny] Grapevine * [Ezy] MLink Assembler Addressmodes ********************** +--------------+------------------------+ | Notation | Description | +--------------+------------------------+ +--------------+------------------------+ | EA | Effective Address | | Dn | D0...D7 | | An | A0...A7 | | Xn | D0...D7, A0...A7 | | .b | Operand Width 8Bit | | .w | Operand Width 16Bit | | .l | Operand Width 32Bit | | size | w,l | | Size | b,w,l | | Scale | 1,2,4 or 8 | | Xn.size*Scale| 68000-10 only Scale 1. | +---------------------------------------+ `Data register direct' Syntax: Dn `Address register direct' Syntax: An `Address register indirect' Syntax: (An) `Address register indirect with postincrement' Syntax: (An)+ `Address register indirect with predecrement' Syntax: -(An) `Address register indirect with offset' Syntax: bd.w(An) `Address register indirect with index and offset' Syntax: bd.b(An,Xn{.Size*Scale}) `Address register indirect with index and offset' Syntax: (bd,An,Xn{.Size*Scale}) `Address register indirect with index and offset' Syntax: (bd.b,An,Xn{.Size*Scale}) `Address register indirect with index and base displacement' Syntax: ({bd.size{,An{,Xn{.Size{*Scale}}}}}) `Indirect Memory Addressierung mit postindex' Syntax: ({[{bd.size{,An}}]}{Xn{.Size{*Scale}{,od.size}}}) `Indirect Memory Addressierung mit preindex' Syntax: ({[{bd.size{,An}}{,Xn{.Size{*Scale}]}{,od.size}}}) `PC Indirect' Syntax: (PC) `PC Indirect with offset' Syntax: bd.w(PC) `PC Indirect with index and offset' Syntax: bd.b(PC,Xn{.Size*Scale}) `PC Indirect with index and offset' Syntax: bd.b(ZPC,Xn{.Size*Scale}) `PC Indirect with index and base displacement' Syntax: ({bd.size{,PC{,Xn{.Size{*Scale}}}}}) `PC Indirect with index and base displacement' Syntax: ({bd.size{,ZPC{,Xn{.Size{*Scale}}}}}) `PC Indirect memory Addressing with post-index' Syntax: ({[{bd.size{,PC}}]}{,Xn{.Size{*Scale}{,od.size}}}) `PC Indirect memory Addressing with post-index' Syntax: ({[{bd.size{,ZPC}}]}{,Xn{.Size{*Scale}{,od.size}}}) `PC Indirect memory Addressing with pre-index' Syntax: ({[{bd.size{,PC}}{,Xn{.Size{*Scale}]}{,od.size}}}) `PC Indirect memory addressing with pre-index' Syntax: ({[{bd.size{,ZPC}}{,Xn{.Size{*Scale}]}{,od.size}}}) `Absolut short' Syntax: bd.w `Absolut long' Syntax: bd[.l] `Immediate Data' Syntax: #xxx Addressmode Examples ==================== To avoid some problems here are some small examples how addressmode have to build up. x=$40 y=$400 move.b (x,A0,D2.W),D0 move.b x(A0,D2.W),D0 ;Both lines are correct ;(x,a0,d2.w) is optimized internal to (x,a0,d2.w). ;For more information please check the chapter about ;Optimizing Direct Addressmodes. move.b (y,A0,D2.W),D0 move.b y(A0,D2.W),D0 ;Now you get 2 errors, because y is not an 8bit word. ;These 2 lines shows the correct version. move.b (y.w,A0,D2.W),D0 move.b (y.w,A0,D2.W),D0 ;or move.b (y.l,A0,D2.W),D0 move.b (y.l,A0,D2.W),D0 680xx Opcode Overview ********************* +----------+---------+-------------------------------------------+ | Opcode | Size | 68000 68010 68020 68030 68040 68060 6888x | +----------+---------+-------------------------------------------+ +----------+---------+-------------------------------------------+ | abcd | b | x x x x x x | | add | b,w,l | x x x x x x | | addq | b,w,l | x x x x x x | | adda | w,l | x x x x x x | | addi | b,w,l | x x x x x x | | addx | b,w,l | x x x x x x | | and | b,w,l | x x x x x x | | andi | b,w,l | x x x x x x | | asr | b,w,l | x x x x x x | | asl | b,w,l | x x x x x x | | bcc | b,w,l | x x x x x x | | bchg | b,l | x x x x x x | | bclr | b,l | x x x x x x | | bfchg | unsized | x x x x | | bfclr | unsized | x x x x | | bfext | unsized | x x x x | | bfffo | unsized | x x x x | | bfins | unsized | x x x x | | bfset | unsized | x x x x | | bftst | unsized | x x x x | | bkpt | unsized | x x x x x | | bset | b,l | x x x x x x | | btst | b,l | x x x x x x | | callm | unsized | x | | cas | b,w,l | x x x x x x,2 | | cas2 | b,w,l | x x x x x x,2 | | chk | b,w,l | x x x x x x | | chk2 | b,w,l | x x x 2 | | cinv¹ | unsized | x x | | clr | b,w,l | x x x x x x | | cmp | b,w,l | x x x x x x | | cmpa | w,l | x x x x x x | | cmpi | b,w,l | x x x x x x | | cmpm | b,w,l | x x x x x x | | cmp2 | b,w,l | x x x 2 | | cpush¹ | unsized | x x | | dbcc | w | x x x x x x | | divs | w,l | x x x x x x,2 | | divsl | l | x x x x | | divu | w,l | x x x x x x,2 | | divul | l | x x x x | | eor | b,w,l | x x x x x x | | eori | b,w,l | x x x x x x | | eori/ccr | b | x x x x x x | | eori/sr¹ | w | x x x x x x | | exg | l | x x x x x x | | ext | w,l | x x x x x x | | extb | l | x x x x | | fabs | | x x x | | fsabs | | x x | | fdabs | | x x | | facos | | 2 2 x | | fadd | | x x x | | fsadd | | x x | | fdadd | | x x | | fasin | | 2 2 x | | fatan | | 2 2 x | | fatanh | | 2 2 x | | fbcc | | x x x | | fcmp | | x x x | | fcos | | 2 2 x | | fcosh | | 2 2 x | | fdbcc | | x 2 x | | fdiv | | x x x | | fsdiv | | x x | | fddiv | | x x | | fetox | | 2 2 x | | fetoxm1 | | 2 2 x | | fgetexp | | 2 2 x | | fgetman | | 2 2 x | | fint | | 2 x x | | fintrz | | 2 x x | | flogn | | 2 2 x | | flognp1 | | 2 2 x | | flog2 | | 2 2 x | | flog10 | | 2 2 x | | fmod | | 2 2 x | | fmove | | x x x | | fsmove | | x x | | fdmove | | x x | | fmovecr | | 2 2 x | | fmovem | | x x,2 x | | fmul | | x x x | | fsmul | | x x | | fdmul | | x x | | fneg | | x x x | | fsneg | | x x | | fdneg | | x x | | fnop | | x x x | | frem | | 2 2 x | | frestore¹| | x x x | | fscc | | 2 2 x | | fsub | | x x x | | fssub | | x x | | fdsub | | x x | | fsave¹ | | x x x | | fscale | | 2 2 x | | fsglmul | | 2 2 x | | fsgldiv | | 2 2 x | | fsin | | 2 2 x | | fsinh | | 2 2 x | | fsincos | | 2 2 x | | fsqrt | | x x x | | fssqrt | | x x | | fdsqrt | | x x | | ftan | | 2 2 x | | ftanh | | 2 2 x | | ftentox | | 2 2 x | | ftrap | | x 2 x | | ftst | | x x x | | ftwotox | | 2 2 x | | illegal | unsized | x x x x x x | | jmp | unsized | x x x x x x | | jsr | unsized | x x x x x x | | lea | l | x x x x x x | | link | w,l | x x x x x x | | lpstop | | x | | lsl | b,w,l | x x x x x x | | lsr | b,w,l | x x x x x x | | move | b,w,l | x x x x x x | | movea | w,l | x x x x x x | | moveq | l | x x x x x x | | movec¹ | l | x x x x x | | movem | w,l | x x x x x x | | movep | w,l | x x x x x | | moves¹ | b,w,l | x x x x x | | move16 | | x x | | muls | w,l | x x x x x x | | mulu | w,l | x x x x x x | | nbcd | b | x x x x x x | | neg | b,w,l | x x x x x x | | negx | b,w,l | x x x x x x | | nop | unsized | x x x x x x | | not | b,w,l | x x x x x x | | or | b,w,l | x x x x x x | | ori | b,w,l | x x x x x x | | pack | unsized | x x x x | | pea | l | x x x x x x | | pflush¹ | unsized | x x x | | pflusha¹ | unsized | x | | plpa¹ | unsized | x | | pload¹ | unsized | x | | pmove¹ | w,l,q | x | | ptest¹ | unsized | x x | | reset¹ | unsized | x x x x x x | | rol | b,w,l | x x x x x x | | ror | b,w,l | x x x x x x | | roxl | b,w,l | x x x x x x | | roxr | b,w,l | x x x x x x | | rtd | unsized | x x x x x | | rte¹ | unsized | x x x x x x | | rtr | unsized | x x x x x x | | rts | unsized | x x x x x x | | rtm | unsized | x | | sbcd | b | x x x x x x | | scc | b | x x x x x x | | stop¹ | unsized | x x x x x x | | sub | b,w,l | x x x x x x | | subq | b,w,l | x x x x x x | | suba | w,l | x x x x x x | | subi | b,w,l | x x x x x x | | subx | b,w,l | x x x x x x | | swap | w | x x x x x x | | tas | b | x x x x x x | | trap | unsized | x x x x x x | | trapcc | ? ,w,l | x x x x | | trapv | unsized | x x x x x x | | tst | b,w,l | x x x x x x | | unlk | unsized | x x x x x x | | unpk | unsized | x x x x | +----------+---------+-------------------------------------------+ ¹ Supervisor instruction 2 These are software-supported instructions on the 68040 and 68060