Probably inaccuracy in GCC 4.7.0 documentation.

[Gerald Pfeifer]

Hi Roman,

On Wed, 16 May 2012, ц╡ц▐ц█ц│ц▌ цЁц│ц√ц┘ц▌ц▀ц▐ц≈ wrote:
> It seems like I've found an inaccuracy in GCC 4.7.0 documentation. It is 
> about -fdefer-pop optimization option. The point is that this option is 
> mentioned in the list of optimization flags which -O1 turns on (Chapter 
> 3: GCC Command Options, part 3.10 Options That Control Optimization, 
> page 107/766 of http://gcc.gnu.org/onlinedocs/gcc-4.7.0/gcc.pdf), but 
> there are no any other allusions to -fdefer-pop in the document, even in 
> 3.1 Option Summary, where it is expected to be. There is no explanation 
> of what this flag does in HTML version of the document too (here 
> [http://gcc.gnu.org/onlinedocs/gcc-4.7.0/gcc/Optimize-Options.html#Optimize-Options] 
> and here 
> [http://gcc.gnu.org/onlinedocs/gcc-4.7.0/gcc/Option-Summary.html#Option-Summary]). 
> I'm guessing that this flag is not further supported by GCC, or not yet 
> implemented.

thank you for reporting this, and sorry for getting back to you
with nearly three years delay.


You did find a real issue in that -fdefer-pop is listed as being
activated by the -O option, but not anywhere else in the documentation.

However, there is actually code to support this option, specifically
in gcc/common.opt and gcc/calls.c:

  calls.c:	else if (flag_defer_pop && inhibit_defer_pop == 0
  common.opt:   Common Report Var(flag_defer_pop) Optimization

I filed a bug report to track this, and hopefully one of the
technical experts is going to address it:

  https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65703

Gerald