public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* Silence overactive sanity check with -fpartial-profile-training
@ 2019-12-08 14:03 Jan Hubicka
  0 siblings, 0 replies; only message in thread
From: Jan Hubicka @ 2019-12-08 14:03 UTC (permalink / raw)
  To: gcc-patches

Hi,
do_estimate_edge_time tests that cached and real values matches. This test
is not working precisely for global profiles because of roundoff issues when
profile of clones is subtracted from profile of offline body.  This is checked
by presence of ipa counter. This breaks with partial profile training because
we turn IPA profiles to local when they drop to 0.

Bootstrapped/regtested x86_64-linux, comitted.

	* ipa-inline-analysis.c (do_estimate_edge_time): Silence overactive
	sanity check.
Index: ipa-inline-analysis.c
===================================================================
--- ipa-inline-analysis.c	(revision 279076)
+++ ipa-inline-analysis.c	(working copy)
@@ -209,6 +209,7 @@ do_estimate_edge_time (struct cgraph_edg
 	  nonspec_time = e->entry.nonspec_time;
 	  hints = e->entry.hints;
 	  if (flag_checking
+	      && !opt_for_fn (callee->decl, flag_profile_partial_training)
 	      && !callee->count.ipa_p ())
 	    {
 	      sreal chk_time, chk_nonspec_time;

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2019-12-08 14:03 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-12-08 14:03 Silence overactive sanity check with -fpartial-profile-training Jan Hubicka

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).