public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
From: Martin Liska <marxin@gcc.gnu.org>
To: gcc-cvs@gcc.gnu.org
Subject: [gcc(refs/users/marxin/heads/loop-unswitch-improvement-v7)] Fix switch bug!
Date: Wed,  8 Dec 2021 11:14:50 +0000 (GMT)	[thread overview]
Message-ID: <20211208111450.984703858410@sourceware.org> (raw)

https://gcc.gnu.org/g:a3ba6f3bb9dff3289345e7b84642c76350502bb9

commit a3ba6f3bb9dff3289345e7b84642c76350502bb9
Author: Martin Liska <mliska@suse.cz>
Date:   Wed Dec 8 12:14:27 2021 +0100

    Fix switch bug!
    
    And add comments.

Diff:
---
 gcc/tree-ssa-loop-unswitch.c | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

diff --git a/gcc/tree-ssa-loop-unswitch.c b/gcc/tree-ssa-loop-unswitch.c
index 811f0cad343..d80b34db861 100644
--- a/gcc/tree-ssa-loop-unswitch.c
+++ b/gcc/tree-ssa-loop-unswitch.c
@@ -90,6 +90,8 @@ struct unswitch_predicate
     edge_index (edge_index_), handled (false)
   {}
 
+  /* Based on true_range, compute inverted range.  */
+
   inline void
   init_false_edge (void)
   {
@@ -100,6 +102,8 @@ struct unswitch_predicate
       false_range.invert ();
   }
 
+  /* Copy ranges for purpose of usage in predicate path.  */
+
   inline void
   copy_merged_ranges ()
   {
@@ -107,11 +111,22 @@ struct unswitch_predicate
     merged_false_range = false_range;
   }
 
+  /* Unswitching expression.  */
   tree condition;
+
+  /* LHS of the expression.  */
   tree lhs;
+
+  /* Initial ranges (when the expression is true/false) for the expression.  */
   int_range_max true_range, false_range;
+
+  /* Modified range that is part of a predicate path.  */
   int_range_max merged_true_range, merged_false_range;
+
+  /* For switch predicates, index of the edge the predicate belongs to.  */
   int edge_index;
+
+  /* True if the predicate was already used for unswitching.  */
   bool handled;
 };
 
@@ -451,6 +466,12 @@ find_unswitching_predicates_for_bb (basic_block bb, class loop *loop,
 		  unswitch_predicate *predicate = new unswitch_predicate (expr, idx, edge_index);
 		  ranger->gori().outgoing_edge_range_p (predicate->true_range, e,
 							idx, *get_global_range_query  ());
+		  /* Huge switches are not supported by Ranger.  */
+		  if (predicate->true_range.undefined_p ())
+		    {
+		      delete predicate;
+		      return;
+		    }
 		  predicate->init_false_edge ();
 
 		  candidates.safe_push (predicate);


             reply	other threads:[~2021-12-08 11:14 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-08 11:14 Martin Liska [this message]
2021-12-08 18:26 Martin Liska
2021-12-09 12:49 Martin Liska

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20211208111450.984703858410@sourceware.org \
    --to=marxin@gcc.gnu.org \
    --cc=gcc-cvs@gcc.gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).