[PATCH] Value profile based optimizations, part 1

[Andreas Jaeger]

Zdenek Dvorak <rakdver@atrey.karlin.mff.cuni.cz> writes:

> static void insn_values_to_profile	PARAMS ((rtx, unsigned *,
> 					    	 struct histogram_value **));

Can you use ISO C 90 prototypes everywhere?  With a new file, we do
not need to use PARAMS or ...
> [...]
> /* Find list of values for that we want to measure histograms.  */
> void
> find_values_to_profile (n_values, values)
>      unsigned *n_values;
>      struct histogram_value **values;

this.
> {
>   rtx insn;
>   unsigned i;
>
>   *n_values = 0;
>   *values = NULL;
>   for (insn = get_insns (); insn; insn = NEXT_INSN (insn))
>     insn_values_to_profile (insn, n_values, values);
>
>   for (i = 0; i < *n_values; i++)
>     {
>       switch ((*values)[i].type)
> 	{
> 	case HIST_TYPE_INTERVAL:
> 	  (*values)[i].n_counters = (*values)[i].hdata.intvl.steps +
> 		  ((*values)[i].hdata.intvl.may_be_less ? 1 : 0) +
> 		  ((*values)[i].hdata.intvl.may_be_more ? 1 : 0);

AFAIR according to the coding styles this should be wrapped around
with the + at the beginning of the line and not at the end.


Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj