Thread: Force 1 byte alignement in 32 bit system

I do not seem able to reproduce your problem with gcc 4.8.4 on Ubuntu 14.04 64-bit:

Code:

#include <stdio.h>
#include <stdint.h>

#define NUMBERS_PER_ROW 20

typedef struct __attribute__((__packed__))
{
    uint8_t tolerance;
    uint8_t footprint[24];
    uint8_t value[24];
    uint8_t comment[100];
    uint8_t type[24];
    uint8_t pitch;
    uint8_t rotation;
} fppStruct_t;
 
typedef union
{
    fppStruct_t fppStruct;
    uint8_t     fppArray[sizeof(fppStruct_t)];
} feederPartParam_t;

int main(void)
{
    printf("%zu %zu\n-------\n", sizeof(fppStruct_t), sizeof(feederPartParam_t));
    feederPartParam_t feeder = {{0}};
    feeder.fppStruct.tolerance = 10;
    feeder.fppStruct.footprint[0] = 1;
    for (size_t i = 0; i < sizeof(fppStruct_t); ++i)
    {
        printf("%2u ", feeder.fppArray[i]);
        if (i % NUMBERS_PER_ROW == NUMBERS_PER_ROW - 1)
        {
            printf("\n");
        }
    }
    printf("\n");
    return 0;
}

when compiled with -std=c99 and run gives the output:

Code:

175 175
-------
10  1  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 
 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 
 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 
 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 
 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 
 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 
 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 
 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 
 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0

Actually, let this be an example for you to follow: notice that instead of saying "feeder is an object of type feederPartParam*", I declared feeder to be a feederPartParam_t object, i.e., I simplified since the pointer was unnecessary: in fact, your statement made no sense as there is no such type as feederPartParam, so there is no such type as feederPartParam*.

Instead of saying that "It skips one byte" while showing code that would have made that happen, I showed code that printed out the array for inspection, and then showed the output that demonstrated that it did not skip one byte.

This way, if I did something contrary to what you did, you can more easily spot it, whereas with your "tell without showing" approach, we have to either assume that you wrote the correct code (in which case why do you need help?), or guess at what went wrong.

I understand. next time I will try to create more showing examples, sorry. I got the values read from debugger.

I am using an arm crosscompiller from Atolic:

Code:

gcc version 4.8.3 20140228 (release) [ARM/embedded-4_8-branch revision 208322] (GNU Tools for ARM Embedded Processors (Build 14.09 Pro))

I will try to work some more on it.

I have used your example with locally created variable and it worked out the same as for you... I am not sure why.

I think I know what is the case... I have an array of sctructs created like this:

Code:

feederObject_t* rs485_feederBlock = 0;
...
size_t blockByteSize = config_globalFlash.flashStruct.nrOfFeeders * sizeof(feederObject_t); // x * 175
rs485_feederBlock = malloc(blockByteSize);
assert_param(rs485_feederBlock);

I fear malloc is making bad alignement, but I am using symbol __packed=__attribute__((__packed__)) so thats weird. feederPartParam_t is one of the members of feederObject_t.