public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "giovannibajo at libero dot it" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug rtl-optimization/15566] Cross-compile ppc64 compile never finishes
Date: Sun, 23 May 2004 23:15:00 -0000	[thread overview]
Message-ID: <20040523032311.27541.qmail@sourceware.org> (raw)
In-Reply-To: <20040521171026.15566.judith@osdl.org>


------- Additional Comments From giovannibajo at libero dot it  2004-05-23 03:23 -------
Confirmed by TM.

The problem appears to be in CSE (log is slightly edited for readability):

[04:43] <tm3> it keeps calling delete_reg_equiv(regno = 122)
[04:43] <tm3> delete_reg_equiv thinks the equivalence is invalid, so it punts 
and just returns
[04:43] <tm3> and it keeps getting called over and over again
[04:44] <pinskia> Law added flush_hash_table in http://gcc.gnu.org/ml/gcc-
patches/1999-04n/msg00240.html
[04:53] <tm3> yah, it looks like flush_hash_table is trying to clear the 
equivalences for reg 22? and going into an infinite loop
[04:54] <tm3> no...hash entry 22
[04:59] <DannyB> the algorithm for memory equivalances is O(n^2)
[05:00] <DannyB> so it just flushes every x insns to keep it fast
[05:02] <DannyB> " /* If we have processed 1,000 insns, flush the hash table to
[05:02] <DannyB>    avoid extreme quadratic behavior."

So, it looks like the underlying algorithm should be fixed instead.

TM: can you please provide the preprocessed testcase that shows this bug? Also, 
which branch did you bootstrap to confirm this?


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |giovannibajo at libero dot
                   |                            |it, tm_gccmail at kloo dot
                   |                            |net, law at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
          Component|c                           |rtl-optimization
     Ever Confirmed|                            |1
           Keywords|                            |compile-time-hog
      Known to fail|                            |3.4.0
   Last reconfirmed|0000-00-00 00:00:00         |2004-05-23 03:23:10
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15566


  parent reply	other threads:[~2004-05-23  3:23 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-05-22 18:09 [Bug c/15566] New: " judith at osdl dot org
2004-05-22 18:19 ` [Bug c/15566] " judith at osdl dot org
2004-05-22 18:19 ` pinskia at gcc dot gnu dot org
2004-05-22 21:44 ` judith at osdl dot org
2004-05-23 23:15 ` giovannibajo at libero dot it [this message]
2004-05-23 23:21 ` [Bug rtl-optimization/15566] CSE delete_reg_equiv inifinite loop (breaks PPC64 bootstrap) giovannibajo at libero dot it
2004-05-26 17:18 ` [Bug rtl-optimization/15566] CSE delete_reg_equiv inifinite loop (breaks PPC64 cross) pinskia at gcc dot gnu dot org
2004-06-21 15:24 ` pinskia at gcc dot gnu dot org

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=20040523032311.27541.qmail@sourceware.org \
    --to=gcc-bugzilla@gcc.gnu.org \
    --cc=gcc-bugs@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).