Re: c++/4235

Re: c++/4235

[Gard, Torbjörn]

The following reply was made to PR c++/4235; it has been noted by GNATS.

From: =?iso-8859-1?Q?=22Gard=2C_Torbj=F6rn=22?=
	 <torbjorn.gard@streamserve.com>
To: "'gcc-gnats@gcc.gnu.org'" <gcc-gnats@gcc.gnu.org>, 
	"'gcc-prs@gcc.gnu.org'" <gcc-prs@gcc.gnu.org>, =?iso-8859-1?Q?=22Gard=2C?=
	=?iso-8859-1?Q?_Torbj=F6rn=22?= <torbjorn.gard@streamserve.com>, 
	"'gcc-bugs@gcc.gnu.org'" <gcc-bugs@gcc.gnu.org>, "'dje@gcc.gnu.org'"
	 <dje@gcc.gnu.org>, "Nilsson, Thomas" <thomas.nilsson@streamserve.com>
Cc:  
Subject: Re: c++/4235
Date: Mon, 10 Sep 2001 10:06:03 +0200

 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=3Dview&pr=3D4235&database=3Dg=
 cc
 
 Hello
 
 > Fix: Install shared libraries (both libgcc_s_*.a and libstdc++.a
 > with chmod a+x,o-r mode so that the shared objects are loaded
 > into the process private segment. GCC exception handling
 > places private data in the library preventing it from being
 > shared.
 
 
 I have tried this fix without success. Either the linker or the loader =
 fails
 to read the libraries.
 
 A possible solution using this method would at best be called a work =
 around
 and
 not a fix. And certainly the PR should not be closed until a real =
 solution
 exists.
 
 It should be noted also that the problem exists in AIX 4.3.3 as well as =
 AIX
 4.2.0.
 
 Please reopen this problem report!
 
 regards
 Torbj=F6rn Gard
 StreamServe

Re: c++/4235

[Nilsson, Thomas]

The following reply was made to PR c++/4235; it has been noted by GNATS.

From: "Nilsson, Thomas" <thomas.nilsson@streamserve.com>
To: gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org, =?iso-8859-1?Q?=22Gard=2C?=
	=?iso-8859-1?Q?_Torbj=F6rn=22?= <torbjorn.gard@streamserve.com>, 
	gcc-bugs@gcc.gnu.org, dje@gcc.gnu.org, "Nilsson, Thomas"
	 <thomas.nilsson@streamserve.com>
Cc:  
Subject: Re: c++/4235
Date: Tue, 11 Sep 2001 10:01:53 +0200

 Hello,
 
 Thanks for looking at the AIX exception bug:
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=4235&database=gcc
 
 I was curious to know why there are two instances of the mutex object in the
 same executable?
 
 Can it really be related to that the libraries are shared by multiple
 processes as your work-around implies?
 
 Regards,
 Thomas