Thoughts on -Winit-self

[Joel Sherrill]

Hi

I was provided a list of warning options for GCC to consider
for future inclusion. I have a local build with all of them added
and was looking to see if the warnings were appropriate and
valid for us.   I would like to get community thoughts on
-Winit-self based on the case it flagged in JFFS2.

I reduced the code to this:

========================
#define uninitialized_var(x) x = x

int jffs2_block_check_erase(void *, void *jeb, unsigned int *bad_offset);

int jffs2_mark_erased_block(void *c, void *jeb)
{
  unsigned int uninitialized_var(bad_offset);

  return jffs2_block_check_erase(c, jeb, &bad_offset);
}
========================

sparc-rtems4.11-gcc -O2 -Wall -Winit-self -c warn.c
warn.c: In function 'jffs2_mark_erased_block':
warn.c:8:34: warning: 'bad_offset' is used uninitialized in this
function [-Wuninitialized]
   unsigned int uninitialized_var(bad_offset);

Essentially this warning is tripped by the programming
idiom used to "inform" the compiler that this variable
is intentionally uninitialized.

The question for the community is:

+ Is this a case we would like to warn about?
+ Is this a programming idiom that is trusted enough that
we should never use this warning?
+ Is there a GCC attribute that we should use instead and
default to this when on a non-GCC compiler?

Basically, do we turn this on or not? If yes, then how to
fix the code warned about. If no, I would like some
rationale.

Thanks.

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel.sherrill at OARcorp.com        On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
Support Available                (256) 722-9985