False positive misra-c2012-9.3

[Emiel Estiévenart]

Hi

On a lot of arrays, I get this warning:
warning GC59630F4: c2012-9.3 MISRA: Required Arrays shall not be partially initialized [misra-c2012-9.3]
Here's an example of such an array:

static t_sensor const * apAllFastHspSensors[] =
{
// #BEGIN_SENSORS
/DEFINE_LIST_CAST (t_sensor const)((void const)/
(t_sensor const)((void const)&entry1),
(t_sensor const)((void const)&entry2),
(t_sensor const)((void const)&entry3),
(t_sensor const)((void const)&entry4),
(t_sensor const)((void const)&entry5),
// #END_SENSORS
(t_sensor const *)(NULL),
};

t_sensor is a typedef of a struct.

I checked by removing the casting and it doesn't change much.
Any advice?

Thanks!

[Daniel Marjamäki]

It might be false positives. However I can't say for sure I fail to reproduce. A small example code that would reproduce the warning would be appreciated.

[Emiel Estiévenart]

Hi,

I was able to reproduce it by creating a typedef of a self referencing struct in a seperate header.
Creating an array of this type causes the error to come up.
When I comment out the reference to self, the error goes away.

How would I fix this?

#ifndef MY_HEADER_H
#define MY_HEADER_H

#include <stdint.h>

typedef struct my_struct3 my_type3_t;

struct my_struct3
{
    my_type3_t* next;
    uint8_t temp;
};

#endif //MY_HEADER_H

[Emiel Estiévenart]

This code in a different source file then throws the error:

static my_type3_t const some_instance =
{
    NULL, 
    0
};

static my_type3_t const* apAllFastHspSensors[] =
{
    (&some_instance),
    (&some_instance),
    (&some_instance),
    (&some_instance),
    (&some_instance),
    NULL,
};

[Daniel Marjamäki]

Thank you. I can reproduce. I believe this is a false positive. In my opinion your code is compliant. I have created this ticket:
https://trac.cppcheck.net/ticket/11707

[Emiel Estiévenart]

Thanks :)

[Daniel Marjamäki]

I have made a fix.
https://github.com/danmar/cppcheck/commit/4cf9a704b4b87ce7387c327705199bd9713c7ee9

If you can test it feel free to ..

[Emiel Estiévenart]

Awesome!
It didn't work at first, but after I removed the cppcheck build output for the offending file the error was gone. Great support, thanks!