Re: [bug report] iio: pressure: bmp280: drop sensor_data array

[Dan Carpenter <dan.carpenter@linaro.org>]

No, I looked at test code from Noah Pendleton it has a mistake.  He's
testing assignments, not initialization.  It's a different thing.

I also looked up the relevant portion from the C11 standard (6.7.9)

  — if it is an aggregate, every member is initialized (recursively)
    according to these rules, and any padding is initialized to zero
    bits;

Which sounds like padding is always zeroed, but the context of this
quote is that we didn't fully set every struct member.  We drop
support for compiler that are over 10 years old so C11 covers
everything we support.

Here is the test code that Noah was using.

https://github.com/memfault/interrupt/blob/master/example/c-struct-padding-initialization/example.c

	struct foo foo;

	// 3. use { 0 } zero-initializer
	memset(&foo, 0xa5, sizeof(foo));
	foo = (struct foo){0};

There isn't an initializer on the foo struct.  Then he does a memset()
and assigns a struct to foo.  Assigning one struct to the other is a
different section of the C standard.

I created my own test:

struct foo {
	unsigned int i;
	unsigned char b;
	// 3 bytes of padding inserted here, UNLESS -fpack-struct is used!
};

static void print_struct(void *buffer, int size)
{
	unsigned char *p = buffer;
	int i;

	for (i = 0; i < sizeof(struct foo); i++) {
		printf("0x%x ", p[i]);
	}
	printf("\n");
}

int main(int argc, char *argv[])
{
	struct foo one = { 0, 0 };
	struct foo two = { 0 };
	struct foo three = { };

	print_struct(&one, sizeof(struct foo));
	print_struct(&two, sizeof(struct foo));
	print_struct(&three, sizeof(struct foo));

	return 0;
}

GCC does not initialize "one" because it's fully defined.  Clang does
initialize it (because Clang goes above and beyond for security).  The
rest are zeroed as expected.

regards,
dan carpenter