bug-binutils
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Bug gas/11395] New: gas creates 64bit add-instruction in 32bit mode for


From: haubi at gentoo dot org
Subject: [Bug gas/11395] New: gas creates 64bit add-instruction in 32bit mode for hppa2.0
Date: 17 Mar 2010 12:57:12 -0000

GNU-as (2.19, 2.20) for hppa2.0w-hp-hpux11.31 creates a different bytecode for
the "ADD" instruction compared to HP-as, when .LEVEL is set to "2.0". There is
no difference when .LEVEL is set to "1.1".

Using this assembly file as input:
$ cat > xx20.s <<EOF
    .LEVEL   2.0
    .CODE
  func
    .PROC
    .CALLINFO   FRAME=0, NO_CALLS
    .EXPORT     func,ENTRY
    .ENTRY
    BV %r0(%rp)
    .EXIT
    ADD     %r1,%r1,%r1  ; <== the important line
    .PROCEND
    .END
EOF
$ /opt/hp-gcc/bin/as xx20.s -o xx20.gnu.o
$ /opt/hp-gcc/bin/objdump -d xx20.gnu.o > xx20.gnu.asm

$ /usr/ccs/bin/as xx20.s -o xx20.hp.o
$ /opt/hp-gcc/bin/objdump -d xx20.hp.o > xx20.hp.asm

$ diff xx20.hp.asm xx20.gnu.asm | grep add
<    4: 08 21 06 01     add r1,r1,r1
>    4: 08 21 06 21     add r1,r1,r1

As there is no cmplt/carry/cond modifier (as described in [1] on page 7-2), I'm
not sure if {The boolean variable "overflow" in the operation section is set}
actually applies.

[1]
http://h21007.www2.hp.com/portal/download/files/unprot/parisc20/PA_7_inst_descriptions.pdf

-- 
           Summary: gas creates 64bit add-instruction in 32bit mode for
                    hppa2.0
           Product: binutils
           Version: 2.20
            Status: NEW
          Severity: normal
          Priority: P2
         Component: gas
        AssignedTo: unassigned at sources dot redhat dot com
        ReportedBy: haubi at gentoo dot org
                CC: bug-binutils at gnu dot org
 GCC build triplet: hppa2.0*-hp-hpux*
  GCC host triplet: hppa2.0*-hp-hpux*
GCC target triplet: hppa2.0*-hp-hpux*


http://sourceware.org/bugzilla/show_bug.cgi?id=11395

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]