2017理财游戏那个最好:DEBUG Tutorial

Note: I do not have a copy of the CLOCK.COM mentioned. If you know where I can find one, by all means, let me know.

                                      DEBUG     This  tutorial is made to present an overview of the DEBUG.COM program forthe IBM PC.   This utility can be extremely useful, when used correctly.  It isalmost  a  must for Assembler Language programmers,  and can  also  provide  aninsight  into  the operation of the machine at the bit level.   It has  severalnice features, including the ability to display and change any of the registersin the IBMPC, start and stop program execution at any time, change the program,and  look  at diskettes,  sector by sector.   DEBUG works at the  machine  codelevel,  but it does also have the ability to disassemble machine code,  and (atdos 2.0), assemble instructions directly into machine code.     The  procedure for starting DEBUG and command syntax will not  be  coveredhere,  as they are well documented in the DOS manual.   What we will do is showsome examples of the various commands and the response which is expected.  Notethat the segment registers will probably not be exactly what is shown.  This isnormal, and should be expected.     For  the examples,  I will be using the demo program CLOCK.COM in the  XA4database.   For  those of you with the IBM assembler (MASM),  the source can bedown loaded.   If you do not have the assembler, or have another assembler, thefile  CLOCK.HEX has been up loaded.   It can be converted to a .COM file  usingany of the existing HEX conversion programs on the SIG.  See the file CLOCK.DOCfor more information.                                STARTING DEBUG     There are two ways to start DEBUG with a file.  Both ways produce the sameresults, and either can be used.     In the Command Line:  A>debug clock.com      Separate from the command line:  A>debug                                       -n clock.com                                      -l     With either method,  you will get the DEBUG prompt of a hyphen (-).  DEBUGhas loaded your program and is ready to run.   The description of each instruc-tion will assume this as a starting point,  unless otherwise mentioned.   If atany time you get different results,  check your procedure carefully.   If it iscorrect,  please leave me a message.   I have tried to check everything,  but Ihave been known to make a mistake or two (anyway).     If you do have problems,  you can enter the command Q (Quit) any time  youhave the DEBUG prompt (-).  This should return you to the DOS prompt.                                RUNNING DEBUG                               DISPLAY COMMANDS                               Register command     The first thing we should look at are the registers,  using the R command.If  you type in an R with no parameters,  the registers should be displayed  asso:AX=0000  BX=0000  CX=0446  DX=0000  SP=FFFE  BP=0000  SI=0000  DI=0000DS=6897  ES=6897  SS=6897  CS=6897  IP=0100   NV UP DI PL NZ NA PE NC6897:0100 E96B01        JMP     026E     CX  contains  the length of the file (0446h or 1094d).   If the file  werelarger  than 64K,  BX would contain the high order of the size.   This is  veryimportant to remember when using the Write command,  as this is the size of thefile to be written.   Remember,  once the file is in memory,  DEBUG has no ideahow large the file is,  or if you may have added to it.   The amount of data tobe written will be taken from the BX and CX registers.     If  we want to change one of the registers,  we enter R and  the  registername.  Let's place 1234 (hexadecimal) in the AX register:     -R AX          R and AX register     AX 0000        Debug responds with register and contents     : 1234         : is the prompt for entering new contents.  We respond 1234     -              Debug is waiting for the next command.     Now if we display the registers, we see the following:AX=1234  BX=0000  CX=0446  DX=0000  SP=FFFE  BP=0000  SI=0000  DI=0000DS=6897  ES=6897  SS=6897  CS=6897  IP=0100   NV UP DI PL NZ NA PE NC6897:0100 E96B01        JMP     026ENote that nothing has changed,  with the exception of the AX register.  The newvalue has been placed in it,  as we requested.  One note.  The Register commandcan only be used for 16 bit registers (AX,  BX,  etc.).  It cannot change the 8bit registers (AH,  AL,  BH,  etc.).  To change just AH, for instance, you mustenter the the data in the AX register, with your new AH and the old AL values.                                 Dump command     One of the other main features of DEBUG is the ability to display areas ofstorage.   Unless you are real good at reading 8088 machine language,  the Dumpcommand is mostly used to display data (text,  flags,  etc.).  To display code,the Unassemble command below is a better choice.   If we enter the Dump commandat this time,  DEBUG will default to the start of the program.   It uses the DSregister as it's default,  and,  since this is a .COM file,  begins at DS:0100.It will by default display 80h (128d) bytes of data, or the length you specify.The  next  execution of the Dump command will display the following 80h  bytes,and so on.   For example, the first execution of D will display DS:0100 for 80hbytes,  the next one DS:0180 for 80h bytes,  etc.   Of course, absolute segmentand  segment register overrides can be used,  but only hex numbers can be  usedfor the offset.  That is, D DS:BX is invalid.     With our program loaded, if we enter the Dump command, we will see this:6897:0100  E9 6B 01 43 4C 4F 43 4B-2E 41 53 4D 43 6F 70 79   ik.CLOCK.ASMCopy6897:0110  72 69 67 68 74 20 28 43-29 20 31 39 38 33 4A 65   right (C) 1983Je6897:0120  72 72 79 20 44 2E 20 53-74 75 63 6B 6C 65 50 75   rry D. StucklePu6897:0130  62 6C 69 63 20 64 6F 6D-61 69 6E 20 73 6F 66 74   blic domain soft6897:0140  77 61 72 65 00 00 00 00-00 00 00 00 00 00 00 00   ware............6897:0150  00 00 00 00 00 00 00 00-00 24 00 00 00 00 00 00   .........$......6897:0160  00 00 00 00 00 00 00 00-00 00 00 00 00 00 00 00   ................6897:0170  00 00 00 00 00 00 00 00-00 00 00 00 44 4F 53 20   ............DOS     Notice that the output from the Dump command is divided into three  parts.On the left, we have the address of the first byte on the line.  This is in theformat Segment:Offset.     Next  comes the  hex data at that location.   Debug will always start  thesecond line at a 16 byte boundary; that is, if you entered D 109, you would get7  bytes of information on the first line (109-10F),  and the second line wouldstart at 110.   The last line of data would have the remaining 9 bytes of data,so 80h bytes are still displayed.     The third area is the ASCII representation of the data.  Only the standardASCII  character set is displayed.   Special characters for the IBMPC  are  notdisplayed;  rather periods (.) are shown in their place.   This makes searchingfor plain text much easier to do.     Dump can be used to display up to 64K bytes of data, with one restriction:It cannot cross a segment boundary.   That is,  D 0100 l f000 is valid (displayDS:0100  to  DS:F0FF),  but  D 9000 l 8000 is not (8000h +9000h  =  11000h  andcrosses a segment boundary).     Since  64K  is 10000h and cannot fit into four hex characters,  Dump  uses0000 to indicate 64K.  To display a complete segment, enter D 0 l 0.  This willdisplay the total 64K segment.     If,  at  any time you want to suspend the display  of  data,  Cntl-NumLockworks as usual.   If you want to terminate the display, Cntl-Break will stop itand return you to the DEBUG prompt.                                    Search     Search  is  used  to find the occurrence of a specific byte or  series  ofbytes  within a segment.   The address parameters are the same as for the  Dumpcommand, so we will not duplicate them here.  However, we also need the data tobe searched for.   This data can be entered as either hexadecimal or  characterdata.   Hexadecimal  data is entered as bytes,  with a space or a comma as  theseparator.   Character  data is enclosed by single or double quotes.   Hex  andcharacter data can be mixed in the same request,  i.e. S 0 l 100 12 34 'abc' 56is  valid,  and requests a search from DS:0000 through DS:00FF for the sequenceof 12h 34h a b c 56h,  in that order.  Upper case characters are different thanlower  case  characters,  and a match will not be found if the  case  does  notmatch.   For  instance,  'ABC'  is not the same as 'abc' or 'Abc' or any  othercombination of upper and lower case characters.  However, 'ABC' is identical to"ABC", since the single and double quotes are separators only.     An example is looking for the string 'Sat'.  Here's what would happen:     -S 0 l 0 'Sat'     6897:0235     -Again,  the  actual segment would be different in your system,  but the  offsetshould be the same.   If we then displayed the data,  we would find the  string'Saturday'  at this location.   We could also search on 'turda',  or any  othercombination  of characters in the string.   If we wanted to find every place wedid an Int 21h (machine code for Int is CD), we would do the following:     -S 0 l 0 cd 21     6897:0050     6897:0274     6897:027F     6897:028B     6897:02AD     6897:02B4     6897:0332     6897:0345     6897:034C     6897:043A     6897:0467     6897:047A     6897:0513     6897:0526     6897:0537     6897:0544     -DEBUG found the hex data CD 21 at the above locations.  This does not mean thatall these addresses are INT 21's, only that that data was there.  It could (andmost likely is) an instruction,  but it could also be an address, the last partof a JMP instruction,  etc.  You will have to manually inspect the code at thatarea to make sure it is an INT 21.   (You don't expect the machine to do every-thing, do you?).                                Compare command     Along  the  same lines of Dump and Search commands,  we have  the  Comparecommand.   Compare  will take two blocks of memory and compare them,  byte  forbyte.  If the two addresses do not contain the same information, both addressesare displayed,  with their respective data bytes.   As an example, we will com-pare DS:0100 with DS:0200 for a length of 8.     -d 0100 l 8 0200     6897:0100  E9  65  6897:0200     6897:0101  6B  70  6897:0201     6897:0102  01  74  6897:0202     6897:0103  43  65  6897:0203     6897:0104  4C  6D  6897:0204     6897:0105  4F  62  6897:0205     6897:0106  43  65  6897:0206     6897:0107  4B  72  6897:0207     None of the eight bytes compared,  so we got output for each byte.   If wehad gotten a match on any of the bytes, DEBUG would have skipped that byte.  ifall of the locations requested matched,  DEBUG would have simply responded withanother prompt.   No other message is displayed.   This is useful for comparingtwo blocks of data from a file,  or a program with the BIOS ROM.   Otherwise, Ihave not found a great amount of use for it.                              Unassemble command     For  debugging,  one of the main commands you will use is  the  Unassemblecommand.   This  command will take machine code and convert it to instructions.Addressing is the same as for previous commands with one exception:   Since  weare  now  working with code (the previous commands are mainly  for  data),  thedefault register is the CS register.  In a .COM program, this makes very littledifference,  unless  you reset the DS register yourself.   However,  in a  .EXEfile,  it  can make a lot of difference,  as the CS and DS registers are set todifferent values.     Unassemble data can lead to some interesting results.   For  instance,  inour example,  CS:IP is set to 6897:0100.   If we look at the program,  we see aJMP as the first instruction,  followed by data.   If we just enter U,  we willstart at CS:IP (6897:0100) and start unassembling data.   What we will get is agood instruction, followed by more or less nonsense.  For instance:     -U     6897:0100 E96B01        JMP     026E     6897:0103 43            INC     BX     6897:0104 4C            DEC     SP     6897:0105 4F            DEC     DI     6897:0106 43            INC     BX     6897:0107 4B            DEC     BXAnd so on,  through 6897:011D.   We know the INC BX, DEC SP, etc. are not validinstructions,  but  DEBUG doesn't,  so we do have to look at the  code.   Afterworking with DEBUG a little, you will be able to spot code versus data with theUnassemble  command.   For now,  suffice to say that the first instruction willtake us to CS:026E and we can start from there.     If we Unassemble CS:026E, we will find something which looks a little morelike what we expect.  We get:     -U 26E     6897:026E 8D167802      LEA     DX,[0278]     6897:0272 B409          MOV     AH,09     6897:0274 CD21          INT     21     6897:0276 EB05          JMP     027D     6897:0278 1B5B32        SBB     BX,[BP+DI+32]     6897:027B 4A            DEC     DX     6897:027C 24B4          AND     AL,B4     6897:027E 30CD          XOR     CH,CL     6897:0280 213C          AND     [SI],DI     6897:0282 027D0A        ADD     BH,[DI+0A]     6897:0285 8D167C01      LEA     DX,[017C]     6897:0289 B409          MOV     AH,09     6897:028B CD21          INT     21     6897:028D CD20          INT     20     The first few instructions look fine.   But,  after the JMP  027D,  thingsstart to look a little funny.  Also, note that there is no instruction startingat 027D.   We have instructions at 027C and 027E,  but not 027D.  This is againbecause DEBUG doesn't know data from instructions.  At 027C, we should (and do)have the end of our data.   But, this also translates into a valid AND instruc-tion,  so DEBUG will treat it as such.   If we wanted the actual instruction at027D,  we could enter U 027D and get it,  but from here,  we don't know what itis.   what I'm trying to say is,  DEBUG will do what ever you tell it.   If youtell it to Unassemble data,  it will do so to the best of its ability.  So, youhave to make sure you have instructions where you think you do.                              DATA ENTRY COMMANDS                                     Enter     The  Enter command is used to place bytes of data in memory.   It has  twomodes:  Display/Modify  and  Replace.   The difference is in where the data  isspecified - in the Enter command itself, or after the prompt.     If you enter E address alone,  you are in display/modify mode.  DEBUG willprompt  you  one  byte at a time,  displaying the current byte  followed  by  aperiod.   At this time,  you have the option of entering one or two hexadecimalcharacters.   If you hit the space bar, DEBUG will not modify the current byte,but  go on to the next byte of data.   If you go too far,  the hyphen (-)  willback up one byte each time it is pressed.     E 103     6897:0103  43.41   4C.42   4F.43   43.     4B.45     6897:0108  2E.46   41.40   53.-     6897:0109  40.47   53.In this example,  we entered E 103.   DEBUG responded with the address and  theinformation  at  that  byte (43).   We entered the 41 and  DEBUG  automaticallyshowed the next byte of data (4C).  Again, we entered 42, debug came back.  Thenext byte was 4F, we changed it to 43.  At 106, 43 was fine with us, so we justhit the space bar.  DEBUG did not change the data, and went on to the followingbytes.  After entering 40 at location 109, we found we had entered a bad value.The  hyphen  key was pressed,  and DEBUG backed up  one  byte,  displaying  theaddress and current contents.  Note that it has changed from the original value(41)  to  the value we typed in (40).   We then type in the correct  value  andterminate by pressing the ENTER key.     As you can see,  this can be very awkward,  especially where large amountsof data are concerned.   Also, if you need ASCII data, you have to look up eachcharacter  and enter its hex value.   Not easy,  to be sure.   That's where theReplace  mod  of operation comes in handy.   Where the Display/Modify  mode  ishandy  for  changing a few bytes at various offsets,  the Replace mode  is  forchanging  several  bytes of information at one time.   Data can be  entered  inhexadecimal or character format,  and multiple bytes can be entered at one timewithout  waiting  for the prompt.   If you wanted to store the  characters  'Myname' followed by a hexadecimal 00 starting at location 103, you would enter:     E 103 'My name' 0As  in  the Search command,  data can be entered in character  (in  quotes)  orhexadecimal  forms  and  can be mixed in the same command.   This is  the  mostuseful way of entering large amounts of data into memory.                                     Fill     The Fill command is useful for storing a lot of data of the same data.  Itdiffers  from  the Enter command in that the list will be  repeated  until  therequested amount of memory is filled.  If the list is longer than the amount ofmemory to be filled,  the extra items are ignored.   Like the Enter command, itwill  take hexadecimal or character data.   Unlike the Enter  command,  though,large amounts of data can be stored without specifying every character.   As anexample, to clear 32K (8000h) of memory to 00h, you only need to enter:     F 0 L 8000 0Which translates into Fill, starting at DS:0000 for a Length of 32K (8000) with00h.   If the data were entered as '1234',  the memory would be filled with therepeating string '123412341234',  etc.   Usually,  it is better to enter  smallamounts of data with the Enter command,  because an error in the length parame-ter of the Fill command can destroy a lot of work.  The Enter command, however,will  only change the number of bytes actually entered,  minimizing the effectsof a parameter error.                                     Move     The Move command does just what it says - it moves data around inside  themachine.   It  takes  bytes from with the starting address and moves it to  theending address.   If you need to add an instruction into a program,  it can  beused  to make room for the instruction.   Beware,  though.   Any data or labelsreferenced after the move will not be in the same place.   Move can be used  tosave a part of the program in free memory while you play with the program,  andrestore  it  at any time.   It can also be used to copy ROM BIOS  into  memory,where it can be written to a file or played with to your heart's content.   Youcan then change things around in BIOS without having to worry about programminga ROM.     M 100 L 200 ES:100This  will  move the data from DS:0100 to DS:02FF (Length 200) to  the  addresspointed to by ES:0100.  Later, if we want to restore the data, we can say:     M ES:100 L 200 100which will move the data back to its starting point.   Unless the data has beenchanged while at the temporary location (ES:0100),  we will restore the data toits original state.                                   AssembleI purposely left the Assemble command to the end,  as it is the most complex ofthe  data entry commands.   It will take the instructions in the assembler lan-guage and convert them to machine code directly.   Some of the things it  can'tdo,  however,  are: reference labels, set equates, use macros, or anything elsewhich cannot be translated to a value.  Data locations have to be referenced bythe physical memory address, segment registers, if different from the defaults,must be specified,  and RET instructions must specify the type (NEAR or FAR) ofreturn to be used.   Also,  if an instruction references data but not registers(i.e.  Mov [278],5), the Byte ptr or Word ptr overrides must be specified.  Oneother  restriction:   To tell DEBUG the difference between moving 1234h into AXand  moving  the data from location 1234 into AX,  the latter is coded  as  MovAX,[1234],  where the brackets indicate the reference is an addressed location.The differences between MASM and DEBUG are as follows:     MASM                DEBUG                    Comments     Mov  AX,1234        Mov  AX,1234             Place 1234 into AX     Mov  AX,L1234       Mov  AX,[1234]           Contents of add. 1234 to AX     Mov  AX,CS:1234     CS:Mov AX,[1234]         Move from offset of CS.     Movs Byte ptr ...   Movesb                   Move byte string     Movs Word ptr ...   Movsw                    Move word string     Ret                 Ret                      Near return     Ret                 Retf                     Far returnAlso,  Jmp instructions will be assembled automatically to Short,  Near, or FarJmps.  However, the Near and Far operands can be used to override the displace-ment if you do need them.  Let's try a very simple routine to clear the screen.     -A 100     6897:0100 mov ax,600     6897:0103 mov cx,0     6897:0106 mov dx,184f     6897:0109 mov bh,07     6897:010B int 10     6897:010D int 20     6897:010F     -     We  are using BIOS interrupt 10h,  which is the video interrupt.  (If  youwould like more information on the interrupt,  there is a very good descriptionin  the Technical Reference Manual.)  We need to call BIOS with  AX=600,  BH=7,CX=0, and DX=184Fh.  First we had to load the registers, which we did at in thefirst  four instructions.   The statement at offset 6897:010B  actually  calledBIOS.   The INT 20 at offset 010D is for safety only.  We really don't need it,but with it in,  the program will stop automatically.   Without the INT 20, andif  we did not stop,  DEBUG would try and execute whatever occurs at 010F.   Ifthis  happens  to  be a valid program (unlikely),  we would  just  execute  theprogram.   Usually,  though,  we will find it to be invalid,  and will probablyhang the system,  requiring a cntl-alt-del (maybe) or a power-off and on  again(usually).  So, be careful and double check your work!     Now, we need to execute the program.  To do this, enter the G command, a Gfollowed  by  the enter key.   If you have entered the program  correctly,  thescreen  will  clear and you will get a message "Program  terminated  normally".(More on the Go command later).     Again, I cannot stress the importance of checking your work when using theAssemble  command.   The commands may assemble correctly,  but cause a  lot  ofproblems.   This  is especially important for the Jmp and Call commands;  sincethey  cause  an interruption in the flow of the program,  they  can  cause  theprogram  to jump into the middle of an instruction,  causing VERY unpredictableresults.                                 I/O commands                                     NameThe  Name  command has just one purpose - specifying the name of a  file  whichDEBUG is going to Load or Write.  It does nothing to change memory or execute aprogram,  but does prepare a file control block for DEBUG to work with.  If youare going to load a program,  you can specify any parameters on the same  line,just  like in DOS.   One difference is,  the extension MUST be specified.   Thedefault is no extension.   DEBUG will load or write any file, but the full filename must be entered.     -n chkdsk.com /fThis statement prepares DEBUG for loading the program CHKDSK.COM passing the /fswitch to the program.   When the Load (see below) command is  executed,  DEBUGwill  load CHKDSK.COM and set up the parameter list (/f) in the program's inputarea.                                     Load     The Load command has two formats.  The first one will load a program whichhas been specified by the Name command into storage, set the various registers,and prepare for execution.   Any program parameters in the Name command will beset into the Program Segment Prefix,  and the program will be ready to run.  Ifthe  file is a .HEX file,  it is assumed to have valid  hexadecimal  charactersrepresenting  memory  values,  two hexadecimal characters per byte.  Files  areloaded  starting at CS:0100 or at the address specified in the  command.    For.COM.  .HEX and .EXE files,  the program will be loaded, the registers set, andCS:IP  set  to the first instruction in the  program.   For  other  files,  theregisters are undetermined, but basically, the segment registers are set to thesegment of the PSP (100h bytes before the code is actually loaded),  and BX andCX are set to the file length.  Other registers are undetermined     -n clock.com     -lThis  sequence  will load clock.com into memory,  set IP to the entry point  of0100,  and CX will contain 0446, the hexadecimal size of the file.  The programis now ready to run.     The second form of the Load command does not use the Name command.   It isused  to load absolute sectors from the disk (hard or soft) into  memory.   Thesector count starts with the first sector of track 0 and continuing to the  endof the track.   The next sector is track 0,  second side (if double sided), andcontinues to the end of that sector.   Then,  back to the first side,  track 1,and so on,  until the end of the disk.   Up to 80h (128d) sectors can be loadedat  one time.   To use,  you must specify starting address,  drive  (0=A,  1=B,etc.), starting sector, and number of sectors to load.     -l 100 0 10 20This instruction tells DEBUG to load, starting at DS:0100, from drive A, sector10h  for 20h sectors.   DEBUG can sometimes be used this way to recover part ofthe  information on a damaged sector.   If you get an error,  check the  memorylocation  for that data.   Often times,  part of the data has been  transferredbefore  the error occurs and the remainder (especially for text files)  can  bemanually entered.   Also,  repetitive retrys will sometimes get the informationinto  memory.   This can then be rewritten on the same diskette (see the  Writecommand below), or copied to the same sector on another diskette.  In this way,the data on a damaged disk can sometimes be recovered.                                     Write     The  write  command is very similar to the Load command.   Both  have  twomodes of operation, and both will operate on files or absolute sectors.  As youhave  probably guessed,  the Write command is the opposite of the Load command.Since all the parameters are the same,  we will not cover the syntax in detail.However,  one  thing worth mentioning:  When using the file mode of  the  Writecommand,  the  amount of data to be written is specified in BX and CX,  with BXcontaining the high-order file size.   The start address can be specified or isdefaulted to CS:0100.   Also, files with an extension of .EXE or .HEX cannot bewritten  out,  and error message to that effect will be displayed.   If you  doneed  to  change a .EXE or .HEX file,  simply rename and  load  it,  make  yourchanges, save it and name it back to its original filename.                                   Input     The  Input command can be used to read a byte of data from any of the  I/Oports  in  the PC.   The port address can be either a one or two byte  address.DEBUG will read the port, and display the contents.     -i 3fd     7D     -This is the Line input port for the first Asynchronous adapter.   Your data maybe different,  as it depends on the current status of the port.   It  indicatesthe  data  in the register at the time it was read was 7Dh.   Depending on  theport, this data may change, as the ports are not controlled by the PC.                                    Output     As you can probably guess,  the Output command is the reverse of the Inputcommand.   You  can use the Output command to send a single byte of data  to  aport.   Note  that certain ports can cause the system to hang (especially thosedealing with system interrupts and the keyboard),  so be careful with what  yousend where!     -o 3fc 1     -Port 3FCh is the modem control register for the first asynchronous port.  Send-ing a 01h to this port turns on the DTR (Data Terminal Ready) bit.   A 00h willturn all the bits off.   If you have a modem which indicates this bit,  you canwatch the light flash as you turn the bit on and off.                              EXECUTION COMMANDS                                      Go     The  Go  command  is used to start program execution.   A  very  versatilecommand, it can be used to start the execution at any point in the program, andoptionally  stop  at any of ten points (breakpoints) in  the  program.   If  nobreakpoints  are set (or the breakpoints are not executed),  program  executioncontinues  until  termination,  in which case the message  "Program  terminatednormally" is sent.   If a breakpoint is executed,  program execution stops, thecurrent registers are displayed, and the DEBUG prompt is displayed.  Any of theDEBUG commands can be executed, including the Go command to continue execution.Note  that the Go command CANNOT be terminated by Cntl-break.   This is one  ofthe few commands which cannot be interrupted while executing.     -g =100The  Go command without breakpoints starts program execution at the address (inthis  case  CS:0100)  in the command.   The equal sign before  the  address  isrequired.   (Without the equal sign, the address is taken as a breakpoint.)  Ifno starting address is specified,  program execution starts at CS:IP.   In thiscase,  since  no breakpoints are specified,  CLOCK.COM will continue  executionuntil the cntl-break key is pressed and the program terminates.   At this time,you will get the message "Program terminated normally".   Note that,  after thetermination  message,  the  program should be reloaded before  being  executed.Also,  any memory alterations (storing data,  etc.) will not be restored unlessthe program is reloaded.     -g 276 47c 528 347This version of the control command will start the program and set  breakpointsat  CS:276,  CS:47C,  CS:528  and  CS:347.   These correspond to  locations  inCLOCK.COM  after the screen is cleared,  and the day,  date and time  are  dis-played,  respectively.   The  program will stop at whichever breakpoint it hitsfirst.   Note  that the second and third breakpoints will only be displayed  attwo times - when the program is started and at midnight.   If you care to  stayup (or just change the time in the computer),  and set a breakpoint at 47C,  twill stop when the program is started, and again at midnight.     Some notes about breakpoints. The execution stops just before the instruc-tion is executed.  Setting a breakpoint at the current instruction address willnot execute any instructions.  DEBUG will set the breakpoint first, then try toexecute the instruction, causing another breakpoint.  Also, the breakpoints useInterrupt  3 to stop execution.   DEBUG intercepts interrupt 3 to stop the pro-gram execution and display the registers.   Finally,  breakpoints are not savedbetween Go commands.  Any breakpoints you want will be have to be set with eachGo command.                                     Trace     Along the same lines as Go is the Trace command.   The difference is that,while Go executes a whole block of code at one time, the Trace command executesinstructions  one at a time,  displaying the registers after each  instruction.Like the Go instruction,  execution can be started at any address.   The  startaddress again must be preceeded by an equal sign.   However,  the Trace commandalso has a parameter to indicate how many instructions are to be executed.     -t =100 5This Trace command will start at CS:100 and execute five instructions.  Withoutthe  address,  execution will start at the current CS:IP value and continue forfive instructions.  T alone will execute one instruction.     When using Trace to follow a program, it is best to go around calls to DOSand interrupts,  as some of the routines involved can be  lengthy.   Also,  DOScannot be Traced,  and doing so has a tendency to hang the system.   Therefore,Trace  to  the call or interrupt and Go to the next address after the  call  orinterrupt.                              ARITHMETIC COMMANDS                                 Hexarithmetic     The  Hexarithmetic command is handy for adding and subtracting hexadecimalnumbers.  It has just two parameters - the two numbers to be added and subtrac-ted.   DEBUG's response is the sum and difference of the numbers.   The numberscan be one to four hexadecimal digits long.   The addition and subtraction  areunsigned, and no carry or borrow is shown beyond the fourth (high order) digit.     -h 5 6     000B FFFF     -h 5678 1234     68AC 4444     -In  the  first example,  we are adding 0005 and 0006.   The sum  is  000B,  thedifference  is -1.   However,  since there is no carry,  we get FFFF.   In  thesecond example, the sum of 5678 and 1234 is 68AC, and the difference is 4444.                                    WRAPUP     If  you give it a chance,  DEBUG can be a very useful tool for the  IBMPC.It  is  almost a requirement for debugging assembler language programs,  as  nonice  error  messages are produced at run time.   DEBUG does work at  the  basemachine  level,  so you need some experience to use it  effectively,  but  withpractice, it will be your most useful assembler language debugging tool.

当程序由DS：100开始执行，那么终止程序时，DEBUG会自动将IP内容重新设定为100。当你要将此程序做成一个独立的可执行文件，则可以用Ｎ命令对该程序命名。但一定要为COM文件，否则无法以DEBUG载入。 

　　 输入N SMILE.COM ；我们得告诉DEBUG程序长度：程序从100开始到106，故占用7 

　　 字节。我们利用BX存放长度值高位部分，而以CX存放低位部分。 

　　 2.输入RBX ；查看 BX 寄存器的内容，本程序只有7个字节，故本步可省略 

　　 3.输入 RCX　 ；查看 CX 寄存器的内容 

　　 4.输入 7　 ；程序的字节数 

　　 5.输入 W ；用Ｗ命令将该程序写入（Write）磁盘中