Hi Ole,
Attached is PA 1.1 zsvjmp.s code generated by gcc. I then hacked the code to
remove
the stack frame and added some comments. See zsvjmp_.s file. The resulting
code is
slightly more optimal than your old assembly fragment.
It assembles okay with current binutils assembler.
Hope this helps,
Dave Anglin
On 2024-09-18 2:31 p.m., Ole Streicher wrote:
Dear m68k enthusiasts,
I am the maintainer of the "iraf" package [1]. Although most of the code is portable, it needs one piece of assembler code for each platform
that provides a (sort of) setjmp for FORTRAN.
A "C" implementation for this looks like:
#include <setjmp.h>
int zsvjmp_( long *buf, long *status ) {
*status = 0;
((long **)buf)[0] = status;
return sigsetjmp ((void *)((long **)buf+1),0);
}
however this does not work, since the "sigsetjmp" call needs to be replaced by
a jump to sigsetjmp instead.
As IRAF is a 40-year old package, there was already some assembler that was
used ~35 years ago [2]:
-------------------------------8<-----------------------------------
.proc
.import setjmp
.export zsvjmp
.callinfo
zsvjmp
;
; save address to status word in jmp_buf[0]
;
stw arg1,0(0,arg0)
ldi 0,1
stws 1,0(0,arg1)
;
; call setjmp with jmp_buf[1]..jmp_buf[51]
;
addi 4,arg0,arg0
b setjmp
nop
;
; setjmp will return directly to the caller of zsvjmp at this
; point, so the next statement will never be reached.
;
nop
.procend
-------------------------------8<-----------------------------------
However, just including this does not work because assembler syntax changed [3].
I created a small repository [4] that contains the assembler I collected
so far as well as two test programs.
As I am not familiar with HP assembler I have no idea how to adopt the assembler code. Maybe someone could help me here? Preferably under the
IRAF license [5], so that it can be included upstream later.
One way that worked for other platforms was to compute the C snipped above (having the sigsetjmp function name by something else) with an
optimizing compiler.
There is no request from the users to have this ported to hppa, and I
doubt that ever will. My main motivation to get it ported is to check
the package for hidden problems on "unusual" architectures.
Best regards
Ole
[1] https://tracker.debian.org/pkg/iraf
[2]
https://github.com/iraf-community/iraf/blob/v2.16.1/unix/portkit/zsvjmp.s.HP800
[3]
https://buildd.debian.org/status/fetch.php?pkg=iraf&arch=hppa&ver=2.18.1%7Erc1-3%7E2&stamp=1726682218&raw=0
[4] https://github.com/olebole/zsvjmp
[5] https://iraf-community.github.io/COPYRIGHT.html
--
John David Anglin [email protected]
.LEVEL 1.1
.text
.align 4
.globl zsvjmp_
.type zsvjmp_, @function
zsvjmp_:
.PROC
.CALLINFO FRAME=64,CALLS,SAVE_RP
.ENTRY
stw %r2,-20(%r30)
copy %r25,%r28
ldo 64(%r30),%r30
stw %r0,0(%r25)
ldo 4(%r26),%r26
ldi 0,%r25
bl __sigsetjmp,%r2
stw %r28,-4(%r26)
ldw -84(%r30),%r2
bv %r0(%r2)
ldo -64(%r30),%r30
.EXIT
.PROCEND
.size zsvjmp_, .-zsvjmp_
.ident "GCC: (Debian 14.2.0-4) 14.2.0"
.section .note.GNU-stack,"",@progbits
.LEVEL 1.1
.text
.align 4
.globl zsvjmp_
.type zsvjmp_, @function
zsvjmp_:
.PROC
; We don't need a frame
.CALLINFO
.ENTRY
copy %r25,%r28
;
; *status = 0
;
stw %r0,0(%r25)
;
; arg0++
;
ldo 4(%r26),%r26
ldi 0,%r25
b __sigsetjmp
;
; ((long **)buf)[0] = status;
;
stw %r28,-4(%r26)
;
; __sigsetjmp will return directly to the caller of zsvjmp at this
; point, so the next statement will never be reached.
;
nop
.EXIT
.PROCEND
.size zsvjmp_, .-zsvjmp_
.ident "GCC: (Debian 14.2.0-4) 14.2.0"
.section .note.GNU-stack,"",@progbits
