Re: FLoating value from native

sharmila inamdar wrote:

>no ,it did not worked.  If I pass float value as 5.5 ad read it back as
>float it comes out to be1.8xxxxxx or something like this. If I give call to
>intbitsToFloat ,then only it works.
>
>What's wrong ?
>
I just hacked this into the NativeLibrary example and it worked fine. 
Here is the relevant output:

original float is 5.5
processed float is 5.5
Exiting Example1

and relevant code fragments:
 From Example1.java:
//...
  public static native float method5(float f);
//...

      float f = 5.5f;
      System.out.println("original float is "+f);
      System.out.println("processed float is "+method5(f));
//...

 From example1*.a51:
;****************************************************************************
;*
;* Function Name: Native_method5
;*
;*   Description: public static native float method5(float f)
;*
;*      Input(s): some float.
;*
;*    Outputs(s): A - zero on success, exception number on failure.
;*                returns a float in R3:R2:R1:R0, msb:...:lsb.
;*               
;*         Notes: Demonstrates processing a float from a native method.
;*
;****************************************************************************
Native_method5:
  clr   a                        ;
  lcall NatLib_LoadPrimitive     ; load a float
                                 ; and simply return it

  clr   a                        ; no error
  ret                            ;
;****************************************************************************


>
>-----Original Message-----
>From: Thomas Chenot <tom.chenot-6tN4nzCoH/[email protected]>
>To: sharmila inamdar <[email protected]>
>Cc: tini <tini-6tN4nzCoH/[email protected]>
>Date: Thursday, April 01, 2004 10:55 PM
>Subject: Re: [TINI]FLoating value from native
>
>
>  
>
>>sharmila inamdar wrote:
>>
>>    
>>
>>>Thanks a lot Shawn and Thomas !
>>>
>>>yes, I got it.  By returning int from native we have to just call
>>>Float.intBitsToFloat() to get the float value from 32bit IEEE754 format.
>>>
>>>      
>>>
>>If you simply declare the return type of the native method to be float,
>>you do not have to call intBitsToFloat. The native method declaration
>>would look something like:
>>
>>public static native float processFloat(float f);
>>
>>You would not have to change anything in your native method implementation.
>>
>>    
>>
>>>Here is a sample java class and a native method.
>>>
>>>
>>>public class FloatRetMath
>>>{
>>> // static method with parameters float and return type int
>>>
>>> public static native int processFloat(float f);
>>>
>>> public static void main(String[] args)
>>> {
>>>
>>> System.loadLibrary("FloatRetMath.tlib");
>>> // attempt to load our native library
>>>float f = 5.0f;
>>>
>>>   try
>>>   {
>>>
>>>    System.out.println("Through Native call passing float " + f );
>>>    int k = processFloat(f);
>>>     System.out.println("Processed Result in integer format is (IEEE754)
>>>      
>>>
>"
>  
>
>>>+ k);
>>>     System.out.println("Processed Result in float is " +
>>>Float.intBitsToFloat(k));
>>>
>>>   }
>>>   catch(Throwable t)
>>>   {
>>>     System.out.println(t);
>>>   }
>>>
>>>   System.out.println("Exiting FloatRetMath");
>>> }
>>>}
>>>
>>>
>>>;*************************************************************************
>>>      
>>>
>**
>  
>
>>>*
>>>Native_processFloat:
>>> clr    a                       ; request first parameter
>>> lcall  NatLib_LoadPrimitive    ; point to parameter
>>>;
>>>; we just popped the param into
>>>; R3:R0 - let's just return them
>>>; as the result
>>>
>>>
>>>
>>>
>>> clr  a                         ; indicate "no exception"
>>> ret                            ;
>>>;*************************************************************
>>>
>>>-----Original Message-----
>>>From: Thomas Chenot <tom.chenot-6tN4nzCoH/[email protected]>
>>>To: sharmila inamdar <[email protected]>
>>>Cc: tini <tini-6tN4nzCoH/[email protected]>
>>>Date: Wednesday, March 31, 2004 9:35 PM
>>>Subject: Re: [TINI]FLoating value from native
>>>
>>>
>>>
>>>
>>>      
>>>
>>>>sharmila inamdar wrote:
>>>>
>>>>
>>>>
>>>>        
>>>>
>>>>>I am using DS80c390. I want to know whether we can get direct floating
>>>>>
>>>>>
>>>>>          
>>>>>
>>>point
>>>
>>>
>>>      
>>>
>>>>>value form native layer ? How should it be returned in native and which
>>>>>format?
>>>>>
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>There is no support for floating point in the native API. The basics
>>>>(+,-,*,/, SQRT) are handled via the JVM. The rest is handled at the
>>>>library level. The format is described in IEEE Std 754-1985. Single
>>>>precision values are 32 bits and are therefore passed in and returned
>>>>just as "int" values are passed and returned. Double precision values
>>>>are 64 bits and therefore are handled in the same manner as "long"
>>>>        
>>>>
>values.
>  
>
>>>>
>>>>        
>>>>
>>>>>The native example given with the TSDK talks about long and int return
>>>>>values. How do I handle float values ? Or I have to return a byte array
>>>>>
>>>>>
>>>>>          
>>>>>
>>>and
>>>
>>>
>>>      
>>>
>>>>>then process it on the JVM layer?
>>>>>
>>>>>
>>>>>..sharmila
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>--
>>>>Thomas Chenot
>>>>
>>>>
>>>>
>>>>        
>>>>
>>>_______________________________________________
>>>TINI mailing list
>>>TINI-6tN4nzCoH/[email protected]
>>>To UNSUBSCRIBE, edit your profile, or see list archives:
>>>http://lists.dalsemi.com/mailman/listinfo/tini
>>>
>>>
>>>
>>>      
>>>
>>--
>>Thomas Chenot
>>
>>    
>>

-- 
Thomas Chenot


_______________________________________________
TINI mailing list
TINI-6tN4nzCoH/[email protected]
To UNSUBSCRIBE, edit your profile, or see list archives:
http://lists.dalsemi.com/mailman/listinfo/tini