public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Jeff Law <law@redhat.com>
To: Peter Bergner <bergner@vnet.ibm.com>
Cc: sje@cup.hp.com, gcc-patches@gcc.gnu.org, paolo.carlini@oracle.com
Subject: Re: Patch to fix gcc.c-torture/compile/20010102-1.c on IA64 HP-UX
Date: Tue, 16 Sep 2008 20:49:00 -0000	[thread overview]
Message-ID: <48D013ED.3050402@redhat.com> (raw)
In-Reply-To: <1221593957.6367.24.camel@sjoa>

Peter Bergner wrote:
> On Tue, 2008-09-16 at 13:14 -0600, Jeff Law wrote:
>   
>> You don't necessarily need to set REG_POINTER on hard registers, and in 
>> general it's unwise.
>>
>> What we do  is we set REG_POINTER on appropriate pseudos within the RTL 
>> for the pseudo.  Then after register allocation/reloading, alter_reg 
>> changes the REGNO field for the pseudo to turn it into a hard register.  
>> Note this preserves the REG_POINTER flag so that you can test it after 
>> reload. 
>>
>> So you'll need to look into where that insn was created and ensure that 
>> the appropriate pseudo has REG_POINTER set (r699)  Then verify that it 
>> was carried forward through allocation/reloading.
>>
>> REG_POINTER, when set, indicates the object is a pointer for its entire 
>> lifetime -- thus setting it on hard registers is generally a bad idea 
>> and it's only safe under certain circumstances on pseudos, particularly 
>> in the presence of cross jumping.
>>     
>
> So something similar to this untested patch to fix Steve's problem?
> (This assumes ORIGINAL_REG (reg) == reg if reg is a psuedo.  Maybe it
> isn't. :)
>
> Peter
>
> --- config/ia64/predicates.md	(revision 140251)
> +++ config/ia64/predicates.md	(working copy)
> @@ -585,6 +585,6 @@ (define_predicate "ar_pfs_reg_operand"
>  (define_predicate "basereg_operand"
>    (match_operand 0 "register_operand")
>  {
> -  return REG_P (op) && REG_POINTER (op);
> +  return REG_P (op) && REG_POINTER (regno_reg_rtx[ORIGINAL_REGNO
> (op)]);
>  })
>
>   
But this shouldn't be necessary.  REG_POINTER is set within the RTX for 
the original pseudo and alter_reg (via changing REGNO) preserves the 
setting for use post-reload, even though we're looking at a hard register.

My recommendation for Steve is the same.  First ensure that REG_POINTER 
is set on the appropriate REG within the insn when the insn is created, 
then go forward and pinpoint where it disappears as the disappearance is 
a bug.
Jeff

  reply	other threads:[~2008-09-16 20:20 UTC|newest]

Thread overview: 51+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-09-16 18:21 Steve Ellcey
2008-09-16 19:00 ` Peter Bergner
2008-09-16 19:20   ` Steve Ellcey
2008-09-16 19:40     ` Peter Bergner
2008-09-16 21:55       ` Jakub Jelinek
2008-09-17  1:22         ` Peter Bergner
2008-09-16 19:44     ` Jeff Law
2008-09-16 20:20       ` Peter Bergner
2008-09-16 20:49         ` Jeff Law [this message]
2008-09-16 20:49           ` Steve Ellcey
2008-09-16 21:31             ` Jeff Law
2008-09-16 21:40               ` Steve Ellcey
2008-09-17  1:54                 ` Peter Bergner
2008-09-17 17:31                   ` Steve Ellcey
2008-09-18 16:03                   ` Steve Ellcey
2008-09-18 20:38                     ` Peter Bergner
2008-09-16 21:48             ` Jeff Law
2008-09-16 22:00               ` Steve Ellcey
2008-09-18 20:59                 ` Richard Henderson
2008-09-19 18:56                   ` Steve Ellcey
2008-09-23 20:55                     ` Jeff Law
2008-09-23 21:08                       ` Steve Ellcey
2008-10-03 19:35                         ` Luis Machado
2008-10-04  0:47                           ` Jeff Law
2008-10-04  1:09                             ` Andrew Pinski
2008-10-16 21:46                               ` Luis Machado
2008-10-16 22:02                                 ` Jeff Law
2008-10-30 22:27                                   ` Luis Machado
2008-10-31  2:23                                     ` Steve Ellcey
2008-10-31  2:17                                       ` Peter Bergner
2008-10-31  2:03                                         ` Jeff Law
2008-10-31  1:50                                           ` Steve Ellcey
2008-11-06 18:00                                             ` Jeff Law
2008-10-31 10:53                                           ` Jakub Jelinek
2008-10-31 20:29                                           ` Peter Bergner
2008-10-31 20:50                                             ` Luis Machado
2008-10-31 21:27                                             ` Jakub Jelinek
2008-11-06 18:25                                     ` Jeff Law
2008-11-06 20:43 David Edelsohn
2008-11-06 22:05 ` Jeff Law
2008-11-06 23:29   ` David Edelsohn
2008-11-06 23:48     ` Jeff Law
2008-11-07 18:58       ` Luis Machado
2008-11-27 14:34       ` Luis Machado
2008-11-27 15:46         ` Richard Guenther
2008-11-27 20:32         ` Jeff Law
2008-11-27 21:07           ` Luis Machado
2008-11-27 23:24             ` Jeff Law
2009-05-26 16:50               ` Luis Machado
2009-06-01  3:55                 ` Ian Lance Taylor
2009-06-01 15:21                   ` Luis Machado

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=48D013ED.3050402@redhat.com \
    --to=law@redhat.com \
    --cc=bergner@vnet.ibm.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=paolo.carlini@oracle.com \
    --cc=sje@cup.hp.com \
    /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).