[Bug c/14005] New: gcc rounds up when it should round down

["lduncan at pillardata dot com" <gcc-bugzilla at gcc dot gnu dot org>]

when trying to round and address down, the compiler rounds up if the sign bit is set on the 
address (and it is marked as signed).

dot-i file that fails:

main ( ) 
{ 
void * pg_ptr1 , * pg_ptr2 ; 
const void * addr1 = ( void * ) 0xfedcba98 ; 
const void * addr2 = ( void * ) 0x1edcba98 ; 

pg_ptr1 = ( void * ) ( ( ( long ) ( addr1 ) / 4096 ) * 4096 ) ; 
pg_ptr2 = ( void * ) ( ( unsigned int ) ( addr1 ) & ~ ( 4096 - 1 ) ) ; 

( void ) printf ( "orig=0x%x, rounding down using val=0x%x\n" , addr1 , 4096 ) ;
 
( void ) printf ( "result: pg_ptr1=0x%p  and pg_ptr2=0x%p => %sequal\n" , 
pg_ptr1 , pg_ptr2 , ( pg_ptr1 == pg_ptr2 ) ? "" : "NOT " ) ; 

pg_ptr1 = ( void * ) ( ( ( long ) ( addr2 ) / 4096 ) * 4096 ) ; 
pg_ptr2 = ( void * ) ( ( unsigned int ) ( addr2 ) & ~ ( 4096 - 1 ) ) ; 

( void ) printf ( "orig=0x%x, rounding down using val=0x%x\n" , addr2 , 4096 ) ;
 
( void ) printf ( "result: pg_ptr1=0x%p  and pg_ptr2=0x%p => %sequal\n" , 
pg_ptr1 , pg_ptr2 , ( pg_ptr1 == pg_ptr2 ) ? "" : "NOT " ) ; 
} 

compiled with "gcc test_prog.c" (i.e. no options)

output from "gcc -v":
Reading specs from /usr/libexec/gcc/darwin/ppc/3.1/specs
Thread model: posix
Apple Computer, Inc. GCC version 1151, based on gcc version 3.1 20020420 (prerelease)

Also found same problem on versions 3.2 (SPARC) and 2.95.3 (QNX cross compiler from SPARC to 
PowerPC)

-- 
           Summary: gcc rounds up when it should round down
           Product: gcc
           Version: 3.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: lduncan at pillardata dot com
                CC: gcc-bugs at gcc dot gnu dot org
  GCC host triplet: PowerPC/Apple Darwin (OS X)


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14005