Re: optimization/6193: [Sparc Solaris2.6] -funroll-loops leads to bootstrap failure in gcc-3.0.4

["Arto Nirkko" <arto.nirkko@insel.ch>]

The following reply was made to PR optimization/6193; it has been noted by GNATS.

From: "Arto Nirkko" <arto.nirkko@insel.ch>
To: bangerth@dealii.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
   arto.nirkko@dkf6.unibe.ch, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: optimization/6193: [Sparc Solaris2.6] -funroll-loops leads to
 bootstrap failure in gcc-3.0.4
Date: Thu, 19 Dec 2002 17:44:46 +0100 (MET)

 Hi
 bug 5888 and 6193 seem resolved since gcc-3.1 and gcc-3.1.1,
 as reported (see http://gcc.gnu.org/ml/gcc/2002-05/msg02737.html).
 About a month ago, you already asked the same question,
 and as a result closed 5888 (see correspondence included below).
 As mentioned in bug 6193, this was a duplicate of 5888,
 which I had to re-submit because of more information available
 (i.e. responsibility of -funroll-loops), since I could't add additional
 info to bug 5888 in gnatsweb (by the way, on other bug-sites, i can add
 additional comments at least for bugs submitted by myself... ;-)
 
 Have not tested gcc-3.2 yet, though (but since 3.1 works and
 hopefully there's no regression...)
 
 Thanks for the query anyway, and good luck
 Arto
 
 
 
 > From anonymous@sources.redhat.com Thu Dec 19 17:21:23 2002
 > Date: Thu, 19 Dec 2002 16:08:57 +0000
 > From: bangerth@dealii.org
 > Subject: Re: optimization/6193: [Sparc Solaris2.6] -funroll-loops leads to
 >  bootstrap failure in gcc-3.0.4
 > To: gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, arto.nirkko@dkf6.unibe.ch,
 >         nobody@gcc.gnu.org
 > Reply-to: bangerth@dealii.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
 >         arto.nirkko@dkf6.unibe.ch, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
 > 
 > Old Synopsis: -funroll-loops leads to bootstrap failure in gcc-3.0.4
 > New Synopsis: [Sparc Solaris2.6] -funroll-loops leads to bootstrap failure in gcc-3.0.4
 > 
 > State-Changed-From-To: open->feedback
 > State-Changed-By: bangerth
 > State-Changed-When: Thu Dec 19 08:08:56 2002
 > State-Changed-Why:
 >     The 3.0 branch has been discontinued. Do you know whether
 >     this works now for gcc3.2 or 3.2.1?
 >     
 >     Thanks
 >       Wolfgang
 > 
 > http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6193
 > 
 --------------old emails---------------
 
 Old Synopsis: Bootstrap stage 2/3 comparison failure gcc-3.0.4
 New Synopsis: [sun solaris 2.6] Bootstrap stage 2/3 comparison failure gcc-3.0.4
 
 State-Changed-From-To: open->feedback
 State-Changed-By: bangerth
 State-Changed-When: Wed Nov 20 17:57:38 2002
 State-Changed-Why:
     This is a rather old bootstrap failure. Can you please check
     whether this still happens with newer versions of gcc and
     report back to us what you find?
     
     Thanks
       Wolfgang
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5888
 
 ---------------------------------------
 Subject: Re: bootstrap/5888: [sun solaris 2.6] Bootstrap stage 2/3 comparison
  failure gcc-3.0.4
 To: bangerth@dealii.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org
 Message-id: <0H5Y00803W4MYK@ubecx01.unibe.ch>
 MIME-version: 1.0
 Content-transfer-encoding: 7BIT
 X-Lines: 41
 
 
 Hi
 no, it doesn't happen anymore with 3.1 and 3.1.1
 (and I have reported it :-) see http://gcc.gnu.org/ml/gcc/2002-05/msg02737.html).
 
 However, I wish to recall my other suggestion again:
 with "gcc -v", the compiler flags used for the bootstrap
 should be reported together with the other configuration
 parameters. This would not only be useful as a requirement
 for bug reporting, but also in "real life" when identifying
 a particular compiler build. Example: I once for test purposes
 bootstrapped with -ffast-math, which naturally leads to certain
 floating point anomalies in programs compiled with the resulting
 compiler. Somehow, I forgot and actually used that compiler,
 leading to errors very difficult to debug, especially since
 gcc -v didn't show anything wrong. Also, bootstrapping options
 like -O3 -mcpu=ultra etc. might be relevent when transporting
 the build to another machine (pre-ultrasparc, for example).
 
 Please let gcc -v report ALL relevant configuration parameters,
 including bootstrapping options!
 Arto
 
 ------------------------------------------
 
 From: bangerth@dealii.org
 Subject: Re: bootstrap/5888: [sun solaris 2.6] Bootstrap stage 2/3 comparison
  failure gcc-3.0.4
 To: gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, arto.nirkko@dkf6.unibe.ch,
         nobody@gcc.gnu.org
 Reply-to: bangerth@dealii.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
         arto.nirkko@dkf6.unibe.ch, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
 Message-id: <20021122150954.28593.qmail@sources.redhat.com>
 
 Synopsis: [sun solaris 2.6] Bootstrap stage 2/3 comparison failure gcc-3.0.4
 
 State-Changed-From-To: feedback->closed
 State-Changed-By: bangerth
 State-Changed-When: Fri Nov 22 07:09:52 2002
 State-Changed-Why:
     Reportedly works now.
     
     Regarding your other suggestion about printing flags on -v,
     please open a new report.
     
     Thanks
       Wolfgang
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5888
 
 -------------------------------------------