Re: target/2246: Code generated for IA64 doesn't reset NaT bit on scratch reg

Re: target/2246: Code generated for IA64 doesn't reset NaT bit on scratch reg

[wilson]

The following reply was made to PR target/2246; it has been noted by GNATS.

From: wilson@gcc.gnu.org
To: gcc-bugs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org,
  sje@cup.hp.com, wilson@gcc.gnu.org
Cc:  
Subject: Re: target/2246: Code generated for IA64 doesn't reset NaT bit on scratch reg
Date: 21 Nov 2001 07:04:47 -0000

 Synopsis: Code generated for IA64 doesn't reset NaT bit on scratch reg
 
 State-Changed-From-To: analyzed->closed
 State-Changed-By: wilson
 State-Changed-When: Tue Nov 20 23:04:46 2001
 State-Changed-Why:
     Andrew MacLeod has contributed two patches to address NaT
     problems.  The first patch fixes the specific problem shown
     by this testcase, and is nearly identical to a patch that
     Steve Ellcey suggested on a private mailing list.  The
     second patch fixes the general instance of this problem
     as described in my comments.
     http://gcc.gnu.org/ml/gcc-bugs/2001-11/msg00501.html
     http://gcc.gnu.org/ml/gcc-bugs/2001-11/msg00500.html
     Both patches have been approved, but have not been
     checked in yet.  That should happen shortly.
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=2246&database=gcc

Re: target/2246: Code generated for IA64 doesn't reset NaT bit on scratch reg

[wilson]

Synopsis: Code generated for IA64 doesn't reset NaT bit on scratch reg

State-Changed-From-To: analyzed->closed
State-Changed-By: wilson
State-Changed-When: Tue Nov 20 23:04:46 2001
State-Changed-Why:
    Andrew MacLeod has contributed two patches to address NaT
    problems.  The first patch fixes the specific problem shown
    by this testcase, and is nearly identical to a patch that
    Steve Ellcey suggested on a private mailing list.  The
    second patch fixes the general instance of this problem
    as described in my comments.
    http://gcc.gnu.org/ml/gcc-bugs/2001-11/msg00501.html
    http://gcc.gnu.org/ml/gcc-bugs/2001-11/msg00500.html
    Both patches have been approved, but have not been
    checked in yet.  That should happen shortly.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=2246&database=gcc

Re: target/2246: Code generated for IA64 doesn't reset NaT bit on scratch reg

[wilson]

Synopsis: Code generated for IA64 doesn't reset NaT bit on scratch reg

Responsible-Changed-From-To: unassigned->wilson
Responsible-Changed-By: wilson
Responsible-Changed-When: Mon Sep 17 21:23:08 2001
Responsible-Changed-Why:
    IA-64 maintainer
State-Changed-From-To: open->analyzed
State-Changed-By: wilson
State-Changed-When: Mon Sep 17 21:23:08 2001
State-Changed-Why:
    The only way to fix this problem is to initialize all
    local variables when they are declared.  Gcc does not
    have infrastructure to do this at the moment, so this
    will require development work.  There is also a concern
    that this will result in redundant stores that the optimizer
    may not be able to eliminate, and thus may result in a
    performance regression.  Additional optimizations may
    be necessary to avoid this problem.
    
    As yet, this has not been observed to cause problems on
    ia64-linux, primarily because gcc does not have speculation
    support, and thus we never set the NAT bit.  This has been
    observed to cause problems on HPUX, as the HPUX compiler
    does use speculation, and thus the NAT bit can be set.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=2246&database=gcc