Assigning an Integer variable to a float variable yields wrong result

Curt Stienstra

Prodigy 60 points

void SomeFunct(...)

{

uint32_t intVal;

float floatVal;

/* some code that sets intVal (to 32 for example) */

floatVal = intVal;

}

When I run this code the floatVal does not = 32.0 but rather 4.484155e-44 (0x00000020 when decoded as a floating point number).

The dis-assembly for this c code is:

1089 floatRpm = rpm;
$C$L137:
00003a42: 9800 LDR R0, [SP]
00003a44: EE000A10 FMSR S0, R0
00003a48: EEB80A40 FUITOS S0, S0

What do I need to do to get the expected floating point val? Here's the basic info about the compiler setup.

over 12 years ago

0 Markel Robregado over 12 years ago

Guru 86765 points

Hi Curt,

Try this below(in bold). See "Type Casting in C".

uint32_t intVal;

float floatVal;

floatVal = (float)intVal;

-kel

0 Chester Gillon over 12 years ago

Guru 92251 points

This could be a result of the debugger giving confusing information when optimisation is used. E.g. when the Optimisation level was set to zero, which allows some intermediate results to be be stored in registers, I was able to repeat the symptoms:

Note that due to optimisation the Variables view shows both the integer intRpm and floating point floatRpm variables in register R0, with the floating point shown with the apparently incorrect value of 4.484155e-44. After stepping out of the test_int_to_float function the value of floatRpm in the caller was correctly shown as 32.0 (in the stack of the caller).

If the test program is re-compiled with the Optimisation level set to Disable then the debugger shows the expected value of the variables, which are now on the stack rather than in registers: