🚀 go-pugleaf

-- 
.ND "10/28/92"
.TL
Bug report
.AU "HA Phung Te" xxx (phungte@cetia.com)
.AF Cetia
.MT ""
.NS
LI, JMC
.NE
.ds HP +2 +1
.de cU  \" commande d'unix
\fC\\$1 \f(CB\\$2\fR
..




.BL
.LI
compiler: 
cc1
.LI
release: 
2.2.2
.LI
target: 
RS6000
.LI 
OS: 
AIX3.2
.LI 
Description of the problem: 
Any file having a function whose name is either \fCexit\fR or \fCabort\fR
can not be compiled.
.LI
Test Case:
.DS
\fCbamboo:/u/phungte/invest/gcc-> cat abort_exit.c
abort(){}
exit(){}
bamboo:/u/phungte/invest/gcc-> gcc -c abort_exit.c
abort_exit.c:1: conflicting types for `abort'
<built-in>:0: previous declaration of `abort'
abort_exit.c:2: conflicting types for `exit'
<built-in>:0: previous declaration of `exit'
bamboo:/u/phungte/invest/gcc-> ls -l abort_exit.o
ls: 0653-341 The file abort_exit.o does not exist.
bamboo:/u/phungte/invest/gcc-> \fR
.DE
.P
.LI 
Cause: 
In \fCinit_decl_processing() (c-decl.c)\fR, \fCexit\fR and \fCabort\fR are
declared as builtin to avoid some spurious warning, according to the comment.
Then later in \fCpushdecl()\fR, the compiler discovers a 
\fCduplicate_decl\fRaration.
.LI 
Temporary Fix: 
The declaration as builtin of these 2 functions (exit and abort) are commented
out. May you tell me if this will hurt the compiler somewhere more than the
fact it will emit spurious warning (I already got a lot of warning, some more
doesn't bother me).