Re: understanding a warning in RTEMS

Sebastian Huber <sebastian.huber@xxxxxxxxxxxxxxxxxx> · Sun, 10 May 2009 12:08:10 +0200

Hi,

you may have a look in section 6.3.1.1 of the C99 standard or search for 
"integer promotions".

CU

Chris Johns wrote:
Hi,

We have this warning in RTEMS (the code is at the end of this email):

$ arm-rtems4.10-gcc -Wall -Wimplicit-function-declaration 
-Wstrict-prototypes -Wnested-externs -fasm -mcpu=arm7tdmi-s 
-mstructure-size-boundary=8 -mapcs-frame -Wextra -Wno-unused 
-Wpointer-arith -Wcast-qual -Wconversion -Wmissing-prototypes -Os -g 
-c -o objwarning.o objwarning.c
objwarning.c: In function 'foo':
objwarning.c:26: warning: conversion to 'uint32_t' from 'int' may 
change the sign of the result

We are using gcc 4.4.0.

I do not understand the reason for the warning and was wondering if 
someone could please help explain it. It happens on all targets I have 
here to test on.

Why does a conversion from the uint16_t argument to a uint32_t only 
generate a warning when we have the extra types used in RTEMS ?

Regards
Chris

The code:

#include <stdint.h>

#define RTEMS_INLINE_ROUTINE static __inline__

typedef uint32_t   Objects_Id;
typedef uint16_t   Objects_Maximum;

#define OBJECTS_INDEX_START_BIT  0U
#define OBJECTS_INDEX_VALID_BITS (Objects_Id)0x0000ffffU

RTEMS_INLINE_ROUTINE Objects_Maximum _Objects_Get_index(
  Objects_Id id
)
{
  return
    (Objects_Maximum)((id >> OBJECTS_INDEX_START_BIT) &
                                          OBJECTS_INDEX_VALID_BITS);
}

RTEMS_INLINE_ROUTINE uint32_t _Objects_Get_index_2(
  Objects_Id id
)
{
  return (id >> OBJECTS_INDEX_START_BIT) & OBJECTS_INDEX_VALID_BITS;
}

uint32_t foo(Objects_Id id, Objects_Id minimum_id, uint16_t v);

uint32_t foo(Objects_Id id, Objects_Id minimum_id, uint16_t v)
{
  uint32_t    block;
  uint32_t    block_2;
  block = _Objects_Get_index( id ) - _Objects_Get_index( minimum_id );
  block_2 = _Objects_Get_index_2( id ) - _Objects_Get_index_2( 
minimum_id );
  return block + block_2;
}