C News Volume 1, Number 2

     Volume 1, Number  2                               10 January 1988
     +---------------------------------------------------------------+
     |                                                               |
     |                                                               |
     |                                                               |
     |                       -  C   News  -                          |
     |                                                               |
     |                                                               |
     |                        International                          |
     |                C Programming & Compiler Review                |
     |                         Newsletter                            |
     |                                                               |
     |                                                               |
     |                                                               |
     +---------------------------------------------------------------+
     US Office:
     Editor                                                Barry Lynch
                                                           Reston, VA
     Foreign Office:   
     Editor                                               David Nugent
          Melbourne, Australia

     C News  is  published  once a month  by  the  C BBS as its official
     newsletter.  You are encouraged to submit articles for publication
     in C News.  Articles should be related to C programming and can be
     Tutorials, reviews or articles of interest to the C programming
     community.  All Operating systems are fairly represented and this
     newsletter shows no favoritism to any one in particular.
   
     C News is the property of the C BBS and is Copyright 1988 by the
     the C BBS.  All rights are reserved and distribution is limited to
     electronic distribution and personal printed copies.  C News cannot
     be resold at any profit, by any organization.  All material enclosed
     within the newsletter is the opinions of the writers and not the
     C BBS or it's Sysop.
   
                             Table of Contents

     1. EDITORIAL
             The Heap: messages from the editor....................  1

     2. BOOK REVIEWS
        C: The Complete Reference
            reviewed by Tom Leonard  ..............................  2
        TurboC Programmer's Reference Book
            reviewed by Tom Leonard  ..............................  2
        Crafting C Tools for the IBM PC.
            reviewed by Mike Shaw  ................................  3

     3. ARTICLES 
        TurboC & TSR's: A few Questions
            by Arnie Cherdak  ......... ...........................  4                   

           C News 1-02                 Page 1                   10 Jan 1988


     =================================================================
                                 EDITORIAL
     =================================================================

     The HEAP: Messages from the Editor.


          Well with only one issue under my belt, the C News -letter is
     already expanding.  David Nugent has voluntered to be a contributing
     editor for future issues of C News.  Since David is located in
     Melbourne, Australia I figured that he could head the foreign office
     there for us.  Gee, the C BBS is now international in scope. 

         Besides, David other C BBS users are pitching in to contribute
     reviews, tutorials and other information to this newsletter.  Three
     users as of this issue have contributed, with others planning to in
     future issues.  Keep in mind that this Newletters letter is for all
     C programmers, and I hope that you will send in questions, articles,
     book reviews and sample programs that you might like to show off.

          Each article has the author's name included in the byline.  Any
     comments on any of the articles should be sent to the those individuals,
     direct if you would like, or to the C BBS. 

          I started an article on the "Legal Aspects of Software Design",
     but decided to get some legal counsel before I put anything on paper.
     Never fear that article will be finished by the next issue.  Speaking
     of which the next issue is scheduled for publication two weeks from
     today - 17 January.  If you would like to submit an article to C News,
     it must be here by the 14th to be included.   Thanks


     B C'ing U..

     Barry
 
           C News 1-02                 Page 2                   10 Jan 1988

     ======================================================================
                                BOOK REVIEWS
     ======================================================================

     ______________________________________________________________________

     Title:     C: The Complete Reference
     Author/s:  Herbert Schildt
     Publisher: Osborne McGraw-Hill
                2600 Tenth Street
                Berkeley, California 94710
     ISBN #:    0-07-881263-1

           If you are just starting into the C langugage you should get a
     copy of this book.  This book tries to be the COMPLETE REFERENCE TO C
     and to a large extent it is.  When I need to known how to use a 
     multidimensional array  or when I want a library routine that will 
     copy a string  this is the book I look in first.

          "C: The Complete Reference" is a large book.  It is over 700 pages
     and it will probably take you a few weeks to read through it but it is
     worth the effort.

     _______________________________________________________________________

     Title:     Turbo C  Programmer's Resource Book
     Author/s:  Fredrick Holtz
     Publisher: TAB books Inc
                P.O. Box 40
                Blue Ridge Summit, PA 17214
     ISBN #:    0-8306-3030-9

            If you are not sure of the differents between a string and a
     character array  or if the following line of codes makes your eyes cross

       int x=0;
       while((a[x] += (a[x]>='A' && a[x]<='Z') ? ' ':0)) ++x;

     Then this books should help quite a bit.  While the title says TURBO C
     the book will be of value with any C compiler.  It very gently and
     carefully explains the C languages.  This is not a complete reference
     to C but is an excellent tutorial for someone just starting into the
     languages.

     Hint:  The above code converts the lower case letters in a string to
            upper case letters.

           C News 1-02                 Page 3                   10 Jan 1988

     ======================================================================
                                BOOK REVIEWS
     ======================================================================

     ______________________________________________________________________
   
     Title:     Crafting C Tools for the IBM PCs
     Author:    Joe Campbell
     Publisher: Prentice-Hall Software Series
     ISBN #:    0-13-188418-2
               
      This book is an excellent one to read if you want to understand  the
      internal workings of the 8086/8088 processor and MS-DOS.  The author
      provides valuable insites into how programs are loaded and  executed
      by DOS and how Assembler-to-C module communication occurs.  Sections
      are  included  on  directory  structure,  keyboard  input  handling,
      standard  output as well as direct video access,  and advanced video
      topics such as coding window subroutines.  The last  topics  covered
      include  coding  your  own  interrupt  handling  routine and  serial
      communication processor. 

      Despite its title, this book is almost one-third Assembler routines. 
      Many Assembler and C routine examples are included; the examples are
      built   up   chapter-to-chapter,    eventually   resulting   in    a
      sophisticated directory listing program. 

      A  section  is  included  on  C  compiler differences,  memory model
      considerations for C programs,  and the  complete  8086  instruction
      set. 

      I recommend this book to anyone who wants to use C and Assembler  to
      write DOS utility programs. 

      _____________________________________________________________________

   

           C News 1-02                 Page 4                   10 Jan 1988

     ======================================================================
     TurboC and TSR's by Arnie Cherdak..
     ======================================================================


     It is unfortunate that Borland's documentation of their INTERRUPT type
     function is rather sketchy.  Certain features appear to be a mystery
     and I wanted to share some of the insights I have had.  My problem
     related to the code generated by the Turbo C compilers.

     INTRODUCTION

     An interrupt service routine (ISR) terminates with an "iret" assembly
     language instruction.  The difference between this and a regular "ret"
     instruction used to return from a regular function call is shown below.

     CPU Action         Instructions>>> ret    far ret      iret
      pop instruction pointer value     YES       Y           Y
      pop code segment value            NO        Y           Y
      pop flags                         N         N           Y

     Functions compiled in the Tiny, Small, and Compact models would use
     the "ret" instruction since they have <= 64K of code.  Medium, Large,
     and Huge model functions would use the far ret instruction to allow
     function calls beyond the 64K limit.

     The values on the stack get there through the normal action brought
     about by the instructions used to get to the functions.  A simple call
     within a 64K segment causes the instruction pointer to be pushed onto
     the stack while a far call causes the code segment value to be added.
     Indirect calls are a bit different but result in the same stack
     configuration.  An interrupt service routine is executed through an
     "int" instruction which causes the flags, segment, and instruction
     pointer to be pushed on the stack before the CPU is sent off to execute
     the routine.

     OBSERVATION

     I recently bought Al Stevens' book, "Turbo C -- Memory-Resident
     Utilities, Screen I/O and Programming Techniques."  In my opinion,
     he has written an exceptionally clear and understandable text and
     has produced a fine bit of programming that works extremely well.
     He has my congratulations on a job well done.

     In going through Stevens' code for interrupt type functions, I saw
     that he invokes ISR's within those functions with simple function
     calls and apparently without accomodation for the iret instruction
     that must appear at the exit point of the ISR.  An example:

           C News 1-02                 Page 5                   10 Jan 1988

     ======================================================================
     TurboC and TSR's by Arnie Cherdak..
     ======================================================================

     /***************************************************************

      QUEST.C -- Test program to examine ISR return code

     ****************************************************************/

     #include <stdio.h>
     #define  intnum 0x16


     void interrupt (*oldisr)(); /* prototype and declaration of oldisr */
                              /* which is a variable declared as a   */
                              /* pointer to an interrupt function.   */


     void interrupt newisr()   /* new isr is executed each time      */
     {                         /* interrupt intnum is executed       */
           (*oldisr)();
     }

     main()                 /* saves old isr vector & installs new */
     {
          oldisr = getvect(intnum);
          setvect(intnum,newisr);
     }

     /******************** End of Quest.c **************************/

     Here, the value of the vector for interrupt, intnum, is retrieved and
     saved and it is then replaced with a vector pointing to the new isr,
     "newisr." OK so far!

     Here's where my understanding goes to pot.  The new isr calls the old
     isr which is a construct Stevens uses quite frequently although there
     is generally other code within the new isr as might be expected.  I left
     it out here to simplify matters.  Now regardless which memory model is
     applied here, it seems that the function call to oldisr should result
     in a far call instruction being pushed onto the stack by Turbo C and,
     as I know from my travels through the BIOS listings, oldisr invariably
     returns with an iret.  It seems that the thing should blow up with the
     stack underflowing almost immediately but it doesn't. It works just
     fine.

           C News 1-02                 Page 6                   10 Jan 1988
     ======================================================================
     TurboC and TSR's by Arnie Cherdak..
     ======================================================================

     Obviously, I missed something.  The Turbo C manuals don't clarify this
     or even seem to mention it.  TC must make an accomodation for the type
     of the called isr vector but I was baffled.  Anybody have any insights
     to share?

     I'm going to compile the above example using the TCC switch to yield
     an assembly language output.  I should be able to see what's happening
     from that.

     /***************** TIME PASSES *********************************/

     THE ANSWER

     Well, I used the "-S" switch on TCC, the command line compiler, and I
     got my answer.  As you can see from the resulting assembly language
     output below, Turbo C inserts a "pushf" instruction ahead of the call
     to "oldisr" which causes the flags to be pushed onto the stack.  Next,
     notice that the call to oldisr is an indirect call using the value of
     the vector to oldisr stored in memory.  In effect, Turbo C has
     circumvented the effects of the compilation memory model type by
     using the pushf and the indirect call which created a stack configuration
     exactly like that produced by an "int" instruction and allowed the "iret"
     in the oldisr to execute without bombing the system.

                           MYSTERY SOLVED!!!

     ;=====================================================================
     ;
     ;      ASSEMBLY LANGUAGE OUTPUT FROM TCC FOR COMPILATION OF
     ;      ISR ROUTINE, QUEST.C
     ;
     ;=====================================================================
                name    quest
     _TEXT   segment byte public 'CODE'
     DGROUP  group   _DATA,_BSS
             assume  cs:_TEXT,ds:DGROUP,ss:DGROUP
     _TEXT   ends
     _DATA   segment word public 'DATA'
     _d@     label   byte
     _DATA   ends
     _BSS    segment word public 'BSS'
     _b@     label   byte
     _BSS    ends
     _TEXT   segment byte public 'CODE'

     ; Line 11
     _newisr proc    far
             push    ax
             push    bx
             push    cx
             push    dx
             push    es
             push    ds
             push    si
             push    di
             push    bp
             mov     ds,cs:DGROUP@

           C News 1-02                 Page 7                   10 Jan 1988

     ======================================================================
     TurboC and TSR's by Arnie Cherdak..
     ======================================================================


     ; Line 12
             pushf                    ; <============== LOOK HERE!!!!!
        call    dword ptr DGROUP:_oldisr  ; <============== Turbo C Did It!
     ; Line 13
     @1:
            pop     bp
            pop     di
            pop     si
            pop     ds
            pop     es
            pop     dx
            pop     cx
            pop     bx
            pop     ax
            iret   
            _newisr endp

      ; Line 16
       _main   proc    near
      ; Line 17
        mov     ax,22
        push    ax
        call    near ptr _getvect
        pop     cx
        cwd   
        mov     word ptr DGROUP:_oldisr+2,dx
        mov     word ptr DGROUP:_oldisr,ax
      ; Line 18
        mov     ax,offset _newisr
        mov     dx,cs
        push    dx
        push    ax
        mov     ax,22
        push    ax
        call    near ptr _setvect
        add     sp,6
      ; Line 19
      @2:
        ret   
        _main   endp
        _TEXT   ends
        _BSS    segment word public 'BSS'
                public  _oldisr
        _oldisr label   dword
                db      4 dup (?)
        _BSS    ends
        _DATA   segment word public 'DATA'
        _s@     label   byte
        _DATA   ends
        _TEXT   segment byte public 'CODE'
                extrn   DGROUP@:word
                extrn   _setvect:near
                public  _newisr
                public  _main
                extrn   _getvect:near
         _TEXT   ends
                 end