Re: optimization/9417: gcc 3.2 on SunOS 5.8 is making broken ucgendat tool under `-m64' and with ANY -O<N> optimization

Re: optimization/9417: gcc 3.2 on SunOS 5.8 is making broken ucgendat tool under `-m64' and with ANY -O<N> optimization

[giovannibajo]

Synopsis: gcc 3.2 on SunOS 5.8 is making broken ucgendat tool under `-m64' and with ANY -O<N> optimization

State-Changed-From-To: feedback->closed
State-Changed-By: bajo
State-Changed-When: Sat May 10 07:51:49 2003
State-Changed-Why:
    No feedback, 3 months.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9417

Re: optimization/9417: gcc 3.2 on SunOS 5.8 is making broken ucgendat tool under `-m64' and with ANY -O<N> optimization

[ehrhardt]

Synopsis: gcc 3.2 on SunOS 5.8 is making broken ucgendat tool under `-m64' and with ANY -O<N> optimization

State-Changed-From-To: open->feedback
State-Changed-By: cae
State-Changed-When: Thu Jan 23 09:31:38 2003
State-Changed-Why:
    Sorry, but this is not really a useful report. Please provide
    some self contained code and show that it is miscompiled by gcc.
    Noone is going to debug a third party application in search for
    a gcc Bug just to find out that that application made an invalid
    assumption that is no longer true with optimizations turned on.
    
    This is just a wild guess but you may try to compile with
    -O3 -fno-strict-aliasing and see if that helps.
    

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9417

optimization/9417: gcc 3.2 on SunOS 5.8 is making broken ucgendat tool under `-m64' and with ANY -O<N> optimization

[mvz]

>Number:         9417
>Category:       optimization
>Synopsis:       gcc 3.2 on SunOS 5.8 is making broken ucgendat tool under `-m64' and with ANY -O<N> optimization
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Jan 23 09:16:01 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Miroslav Zubcic
>Release:        Reading specs from /usr/local/lib/gcc-lib/sparc-sun-solaris2.8/3.2/specsConfigured with: ../configure --with-as=/usr/ccs/bin/as --with-ld=/usr/ccs/bin/ld --disable-nls Thread model: posix gcc version 3.2
>Organization:
>Environment:
SunOS 5.8 (Solaris 8)
SunOS zvijer 5.8 Generic_108528-17 sun4u sparc SUNW,Ultra-Enterprise
>Description:
While compiling OpenLDAP 2.1.12 (http://www.OpenLDAP.org)
gcc 3.2 is making broken one small peace of code. I have reported bug on openldap-bugs list and OpenLDAP developer
told me that this look like optimization bug uin gcc 3.2

The mentioned code is in penldap-2.1.12/libldap/liblunicode.
gcc is making help tool "ucgendat" which is producing unicode ".dat" files for use by libldap library, slapd(8) daemon, slapindex(8) and slapadd(8).

If I use -O3 -O2 or -O (as for the rest of OpenLDAP code)
ucgendat will produce broken unicode tables in dat files.
As a consequence mentioned programs will segfault while
indexing LDAP attributes which have UTF-8 non/ascii letters
in their values.

After recompiling ucgendat without optimization and making
dat files with *that* ucgendat, everything seems to be OK.
>How-To-Repeat:
Compile OpenLDAP on SunOS 5.8, with `-m64 -O3 -mcpu=ultrasparc"

install it, make a small LDAP database, put non-ascii values in cn or sn attribute, write

"index           cn,sn,uid   pres,eq,approx,sub"

in slapd.conf(5) after database definition and run slapindex(8)
tool.
>Fix:
Dunno, For now, I compile unicode dat files again after
optimized OpenLDAP build as a workaround.
>Release-Note:
>Audit-Trail:
>Unformatted: