CCS C Software and Maintenance Offers
FAQFAQ   FAQForum Help   FAQOfficial CCS Support   SearchSearch  RegisterRegister 

ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

CCS does not monitor this forum on a regular basis.

Please do not post bug reports on this forum. Send them to support@ccsinfo.com

if/else vs switch/case

 
Post new topic   Reply to topic    CCS Forum Index -> General CCS C Discussion
View previous topic :: View next topic  
Author Message
Reler
Guest







if/else vs switch/case
PostPosted: Fri Mar 05, 2004 5:36 am     Reply with quote

Hi all.
I have a strange problem with my code.
This function is compiled ok but it does not work right.

Code:

void UnlockCom(eComDev Device)
{
   switch(Device)
   {
      case COMDEV_BT:
      case COMDEV_USB:
      case COMDEV_RADIO:         
          bCom1Lock = false;
          break;
         
      case COMDEV_GSM:
      case COMDEV_BAL:
         bCom2Lock = false;
         break;
   }
}


If I call it with COMDEV_BAL for example, bCom2Lock is not set to false.
Instead, this one runs well:

Code:

void UnlockCom(eComDev Device)
{
   if(Device==COMDEV_BT || Device==COMDEV_USB || Device==COMDEV_RADIO)
   {
      bCom1Lock=false;
      return;
   }
   if(Device==COMDEV_GSM || Device==COMDEV_BAL)
      bCom2Lock = false;
}


What's happening? I don't understand it.
eComDevice is an enumeration:

Code:

enum  eComDev {COMDEV_NONE,COMDEV_USB,COMDEV_BT,COMDEV_RADIO,COMDEV_GSM,COMDEV_BAL};


I am using PCH 3.184
Thanks in advance.
mpfj



Joined: 09 Sep 2003
Posts: 95
Location: UK

View user's profile Send private message Send e-mail Visit poster's website

PostPosted: Fri Mar 05, 2004 8:04 am     Reply with quote

Could you post the listing file (or relevant portion) ?

Mark
Charlie U



Joined: 09 Sep 2003
Posts: 183
Location: Somewhere under water in the Great Lakes

View user's profile Send private message

PostPosted: Fri Mar 05, 2004 8:29 am     Reply with quote

First, you have not included all of the possible enumeration values and they are not in enumerated order. This may not be a problem, but I try to do it when I have a switch with an enumerated variable.

You also have 2 lines following the COMDEV_RADIO case. I think the compiler is performing the bCom1Lock = false; line as you intended, in response to the switch, then just executing the break and never making it to the final 2 case statements. (See the added braces)

Try this and see if it helps:

Code:

void UnlockCom(eComDev Device)
{
   switch(Device)
   {
      case COMDEV_NONE:
         break;
      case COMDEV_USB:
      case COMDEV_BT:
      case COMDEV_RADIO:
      {
          bCom1Lock = false;
          break;
      }
         
      case COMDEV_GSM:
      case COMDEV_BAL:
      {
         bCom2Lock = false;
         break;
      }
   }
}
mpfj



Joined: 09 Sep 2003
Posts: 95
Location: UK

View user's profile Send private message Send e-mail Visit poster's website

PostPosted: Fri Mar 05, 2004 8:41 am     Reply with quote

Argh ... those pesky brackets !! I can't believe I missed that one ...
RKnapp



Joined: 23 Feb 2004
Posts: 51

View user's profile Send private message

PostPosted: Fri Mar 05, 2004 9:31 am     Reply with quote

But those brackets shouldn't be necessary. The initial function looked fine! In fact, the inclusion of those brackets would imply that the COMDEV_NONE case would need them, too.

To me, it looks as if something funny is happening with the compiler's interpretation of the enumeration. Do the values of the enum start with 0 and increment? Are they int8s? It seems possible to me that all the values of the enum wind up as zero, so that in the initial form of this function, which looks fine to me, calling it with anything unlocks Com1 but never Com2.

Just a thought,

Smile Robert
mpfj



Joined: 09 Sep 2003
Posts: 95
Location: UK

View user's profile Send private message Send e-mail Visit poster's website

PostPosted: Fri Mar 05, 2004 9:45 am     Reply with quote

Agreed, the brackets *shouldn't* be necessary, but I'd like to see the listing file all the same ...
mpfj



Joined: 09 Sep 2003
Posts: 95
Location: UK

View user's profile Send private message Send e-mail Visit poster's website

Re: if/else vs switch/case
PostPosted: Fri Mar 05, 2004 10:00 am     Reply with quote

I've compiled a copy of your code and, looking at the listing, the compiler is using a table lookup routine for the first piece of code, and "normal" compare / jump code for the second.

What version of the compiler are you using ? Some versions have problems with table loopkups.

If you add a "default" statement (good programming practice I was always told ??), the compiler switches back to compare / jump code. Try it, and see if it makes a difference ...
Reler
Guest







PostPosted: Fri Mar 05, 2004 1:53 pm     Reply with quote

Hi again and thanks for your help.
Here is the code generated by the compiler for both functions.

Code:

.................... void UnlockCom(eComDev Device)
.................... {
....................    if(Device==COMDEV_BT || Device==COMDEV_USB || Device==COMDEV_RADIO)
....................    {
*
046C:  MOVLB  A
046E:  MOVF   x42,W
0470:  SUBLW  02
0472:  BTFSC  FD8.2
0474:  GOTO   048C
0478:  DECFSZ x42,W
047A:  GOTO   0482
047E:  GOTO   048C
0482:  MOVF   x42,W
0484:  SUBLW  03
0486:  BTFSS  FD8.2
0488:  GOTO   0498
....................       bCom1Lock=false;
048C:  MOVLB  0
048E:  BCF    x77.1
....................       Com1LockedDev=COMDEV_NONE;
0490:  CLRF   x79
....................       return;
0492:  GOTO   04B2
0496:  MOVLB  A
....................    }
....................    if(Device==COMDEV_GSM || Device==COMDEV_BAL)
....................    {
0498:  MOVF   x42,W
049A:  SUBLW  04
049C:  BTFSC  FD8.2
049E:  GOTO   04AC
04A2:  MOVF   x42,W
04A4:  SUBLW  05
04A6:  BTFSS  FD8.2
04A8:  GOTO   04B4
....................       bCom2Lock = false;
04AC:  MOVLB  0
04AE:  BCF    x77.5
....................       Com1LockedDev=COMDEV_NONE;
04B0:  CLRF   x79
04B2:  MOVLB  A
04B4:  MOVLB  0
04B6:  RETLW  00


Code:

.................... void UnlockCom(eComDev Device)
.................... { 
....................    switch(Device)
....................    {
*
046C:  MOVLW  01
046E:  MOVLB  A
0470:  SUBWF  x42,W
0472:  ADDLW  FB
0474:  BTFSC  FD8.0
0476:  GOTO   0494
047A:  ADDLW  05
047C:  MOVLB  0
047E:  GOTO   0498
....................       case COMDEV_BT:
....................       case COMDEV_USB:
....................       case COMDEV_RADIO:
....................          bCom1Lock = false;
0482:  BCF    x77.1
....................          break;
0484:  MOVLB  A
0486:  GOTO   0494
....................
....................       case COMDEV_GSM:
....................       case COMDEV_BAL:
....................         bCom2Lock = false;
048A:  BCF    x77.5
....................          break;
048C:  MOVLB  A
048E:  GOTO   0494
0492:  MOVLB  A
....................    }
.................... }
0494:  MOVLB  0
0496:  RETLW  00


There is something strange (at least for me) in the second one. A jump in 047E to 0498 out of the scope of the function. In any case, I have searched for that address but it's not in the list file... ¿?

As I wrote before, compiler version is PCH 3.184
Guest








PostPosted: Fri Mar 05, 2004 2:42 pm     Reply with quote

Very interesting... Like mpfj said, adding a default stament code changes completely switching to "compare & jump mode".

This is the new code generated:

Code:

.................... void UnlockCom(eComDev Device)
.................... {
....................    /*
....................    if(Device==COMDEV_BT || Device==COMDEV_USB || Device==COMDEV_RADIO)
....................    {
....................       bCom1Lock=false;
....................       Com1LockedDev=COMDEV_NONE;
....................       return;
....................    }
....................    if(Device==COMDEV_GSM || Device==COMDEV_BAL)
....................    {
....................       bCom2Lock = false;
....................       Com1LockedDev=COMDEV_NONE;
....................    }
....................    */
....................    switch(Device)
....................    {
*
046C:  MOVLB  A
046E:  MOVF   x42,W
0470:  MOVWF  00
0472:  MOVLW  02
0474:  SUBWF  00,W
0476:  MOVLB  0
0478:  BTFSC  FD8.2
047A:  GOTO   04AA
047E:  MOVLW  01
0480:  SUBWF  00,W
0482:  BTFSC  FD8.2
0484:  GOTO   04AA
0488:  MOVLW  03
048A:  SUBWF  00,W
048C:  BTFSC  FD8.2
048E:  GOTO   04AA
0492:  MOVLW  04
0494:  SUBWF  00,W
0496:  BTFSC  FD8.2
0498:  GOTO   04B0
049C:  MOVLW  05
049E:  SUBWF  00,W
04A0:  BTFSC  FD8.2
04A2:  GOTO   04B0
04A6:  GOTO   04B6
....................       case COMDEV_BT:
....................       case COMDEV_USB:
....................       case COMDEV_RADIO:
....................          bCom1Lock = false;
04AA:  BCF    x77.1
....................          break;
04AC:  GOTO   04B6
....................
....................       case COMDEV_GSM:
....................       case COMDEV_BAL:
....................         bCom2Lock = false;
04B0:  BCF    x77.5
....................          break;
04B2:  GOTO   04B6
....................
....................       default:
....................    }
.................... }
04B6:  RETLW  00


At this moment I don't have the hardware with me to test it, but it looks better.

I know default statment is optional in ANSI C, so I have checked ccs-manual to see if it is a compiler limitation, but I have found this:

Code:

STATEMENTS
switch (expr)  {
case cexpr: stmt; //one or more case [default:stmt]

... }
Note:   Items in [  ] are optional


So, must I double check all the list file looking for more thinks like this before releasing my application?
Ttelmah
Guest







PostPosted: Fri Mar 05, 2004 4:03 pm     Reply with quote

The change in compiler 'mode' for switches, is something I have posted about on many occasions. It will normally default to using the 'compare' mode for smaller switches, and the 'boundary' is usually about 5 tests. Over this number of tests, the compiler uses the jump table, unless a 'default' is present. Hence, if you want to use the table for a switch with effectively a default (for speed), you have to code something like:
Code:

if (test<1 || test>6) {
   //default code
}
else
   switch test {
   case 1:

   case 2:

   case 3:

   case 4:

   case 5:

   case 6:

   }


Several people in the past have suggested, that the behaviour would be much nicer if it was controllable with a compiler 'switch', (something like #use tableswitch), especially since the presence of a jump table, can cause problems on some chips, and increases the number of registers that have to be saved when entering the interrupt routines...
Normally the jump table code is now fine. It does appear that the use of the enumeration, may be causing an unexpected behaviour in this regard, and really should be reported to CCS.

Best Wishes
Display posts from previous:   
Post new topic   Reply to topic    CCS Forum Index -> General CCS C Discussion All times are GMT - 6 Hours
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group