Re: port-hp300/50852: src/sys/arch/hp300/stand/common/ite.c:239: bad expression

[Dave Huang <khym%azeotrope.org@localhost>]

On Fri, Feb 26, 2016 at 09:24:14PM +0100, Joerg Sonnenberger wrote:
> On Fri, Feb 26, 2016 at 04:45:21PM +0000, David Binderman wrote:
> > >> Source code is
> > >>
> > >> whichconsole = ++whichconsole % (NITE+1);
> > >>
> > >> Maybe better code
> > >>
> > >> whichconsole = (whichconsole + 1) % (NITE+1);
> > >
> > > Independent of the question of which code is better, I don't think the
> > > error is correct. 
> > 
> > False, the error is valid.
> > 
> > There are two writes to whichconsole, one on the left of the assignment,
> > one on the right.
> > 
> > The ISO standard doesn't define which order those writes occur.
> > See Steve Summit's C FAQ for sequence points, section 3.8.
> 
> You are wrong. There is no write from the left operand of the
> assignment. The write is the result of the assignment and as written
> before is explicitly sequenced by the standard.

How is the code in question different from i = ++i;, which I've always
been told is undefined? The assignment operator does not introduce a
sequence point.

gcc thinks it's wrong too:

% cat t.c
#define NITE 4

int main(void)
{
  int whichconsole = 0;
  whichconsole = ++whichconsole % (NITE+1);
  return 0;
}

% cc -Wall t.c
t.c: In function ‘main’:
t.c:6:16: warning: operation on ‘whichconsole’ may be undefined [-Wsequence-point]
   whichconsole = ++whichconsole % (NITE+1);
                ^