On Thu, Dec 10, 2020 at 03:18:54PM +0100, Eric Botcazou wrote:
> OK, thanks, but aren't there missing TABs in the new version? Only one line
> is changed in the end AFAICS.
No, sorry, just me hand editing of the patch (and the source) instead of
regenerating the patch (I had there prob.invert (); instead of prob =
prob.invert (); before).
Here is the updated patch that shows that only the 1 -> 2 lines changed
appart from the comments.
2020-12-10 Jakub Jelinek <ja...@redhat.com>
PR rtl-optimization/98212
* dojump.c (do_compare_rtx_and_jump): Change computation of
first_prob for and_them. Add comment explaining and_them case.
* gcc.dg/predict-8.c: Adjust expected probability.
--- gcc/dojump.c.jj 2020-12-10 12:30:50.677948803 +0100
+++ gcc/dojump.c 2020-12-10 15:40:01.577251502 +0100
@@ -1138,18 +1138,37 @@ do_compare_rtx_and_jump (rtx op0, rtx op
cprob = cprob.apply_scale (99, 100);
else
cprob = profile_probability::even ();
- /* We want to split:
+ /* For and_them we want to split:
if (x) goto t; // prob;
+ goto f;
into
- if (a) goto t; // first_prob;
- if (b) goto t; // prob;
+ if (a) ; else goto f; // first_prob for ;
+ // 1 - first_prob for goto f;
+ if (b) goto t; // adjusted prob;
+ goto f;
such that the overall probability of jumping to t
- remains the same and first_prob is prob * cprob. */
+ remains the same. The and_them case should be
+ probability-wise equivalent to the !and_them case with
+ f and t swapped and also the conditions inverted, i.e.
+ if (!a) goto f;
+ if (!b) goto f;
+ goto t;
+ where the overall probability of jumping to f is
+ 1 - prob (thus the first prob.invert () below).
+ cprob.invert () is because the a condition is inverted,
+ so if it was originally ORDERED, !a is UNORDERED and
+ thus should be relative 1% rather than 99%.
+ The invert () on assignment to first_prob is because
+ first_prob represents the probability of fallthru,
+ rather than goto f. And the last prob.invert () is
+ because the adjusted prob represents the probability of
+ jumping to t rather than to f. */
if (and_them)
{
rtx_code_label *dest_label;
prob = prob.invert ();
- profile_probability first_prob = prob.split (cprob).invert ();
+ profile_probability first_prob
+ = prob.split (cprob.invert ()).invert ();
prob = prob.invert ();
/* If we only jump if true, just bypass the second jump. */
if (! if_false_label)
@@ -1163,6 +1182,15 @@ do_compare_rtx_and_jump (rtx op0, rtx op
do_compare_rtx_and_jump (op0, op1, first_code, unsignedp,
mode,
size, dest_label, NULL, first_prob);
}
+ /* For !and_them we want to split:
+ if (x) goto t; // prob;
+ goto f;
+ into
+ if (a) goto t; // first_prob;
+ if (b) goto t; // adjusted prob;
+ goto f;
+ such that the overall probability of jumping to t
+ remains the same and first_prob is prob * cprob. */
else
{
profile_probability first_prob = prob.split (cprob);
--- gcc/testsuite/gcc.dg/predict-8.c.jj 2020-12-10 12:30:50.677948803 +0100
+++ gcc/testsuite/gcc.dg/predict-8.c 2020-12-10 12:31:06.639772968 +0100
@@ -8,4 +8,4 @@ int foo(float a, float b) {
return 2;
}
-/* { dg-final { scan-rtl-dump-times "65.\[34]. .guessed" 2 "expand"} } */
+/* { dg-final { scan-rtl-dump-times "99.\[678]. .guessed" 2 "expand"} } */
Jakub
