PATCH: Fix toplev.c breakage on PA after eh merge

[John David Anglin]

> >> They are also being output before it has been
> >> determined whether or not they are in fact needed.  If this is fixed,
> >> I believe that the patch to open the file earlier shouldn't be needed.
> 
> > Doesn't matter.  At least on the PA, it's perfectly fine to import
> > something you never use.  We've been doing that for a long long time, so
> > I suspect it's OK on other platforms too.
> 
> It seems from semi-recent testsuite bug reports that an import directive
> produces an undefined symbol in the assembler output, which might cause
> something to be linked in that is never actually used.  FWIW.

I have one instance of this problem under vax-ultrix or at least something
similar when I try to build libstdc++-v3.  The undefined symbol is output
in a .stabs directive.  I have found that removing the two directives
allows me to build v3.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

--- misc-inst.s.orig	Sun Apr 22 17:13:23 2001
+++ misc-inst.s	Sun Apr 22 17:45:09 2001
@@ -41589,7 +41589,6 @@
 	.stabs	"__cerb:1282",128,0,1169,-2
 	.stabn	192,0,0,LBB3073
 	.data
-	.stabs	"__n:V28",38,0,1173,__ZNSs4_Rep11_S_max_sizeE
 	.stabs	"__idelim:r1",64,0,1175,7
 	.stabs	"__sb:r1121",64,0,1176,6
 	.stabs	"__c:r1",64,0,1177,2
@@ -41608,7 +41607,6 @@
 	.stabn	192,0,0,LBB3117
 	.stabn	224,0,0,LBE3117
 	.stabn	224,0,0,LBE3116
-	.stabs	"__n:V28",38,0,1173,__ZNSs4_Rep11_S_max_sizeE
 	.stabs	"__idelim:r1",64,0,1175,7
 	.stabs	"__sb:r1121",64,0,1176,6
 	.stabs	"__c:r1",64,0,1177,2