|
HP OpenVMS Systemsask the wizard |
|
The Question is: I'm in the process of converting some MACRO-32 code from VAX to ALPHA. The MACRO is a subroutine that gets called from FORTRAN programs. The subroutine takes 9 parameters. The original code made use of references to the AP. This was causing me some grief in this case, although in other routines with fewer parameters, I had no issues. I've decided to alter the code to conform to the ALPHA calling standards. I replaced references to (AP) for the argument count to R25. The first six parameters, I replaced with the appropriate registers (R16 thru R21). According to table 3-14 in the "OpenVMS Calling Standard" manual, the remaining 3 parameters should be referenced as 0(SP), 8(SP) and 16(SP), respectively. When I did this, the compiler complained with an error of UPLEVSTK. I'm not sure how why or how to get around this. Can you provide an example of MACRO subroutine that takes more than 6 parameters that I can refer to? Thanks in advance for any help you can provide. Tim DiBendetto 610-407-6181 The Answer is :
First, determine if you even need the Macro32 code. In many
cases, the Macro32 code is no longer required and can be
replaced as alternative solution(s) are now available.
Then make only the recommended motifications to the Macro32 code,
there is no need to modify the register operations to match the
Macro64 assembler registers as you are working with the Macro32
compiler and not the Macro64 assembler.
The Macro32 compiler will correctly process argument lists with
more than six arguments -- the code generated by the Macro32
compiler will correctly and transparently use the six Alpha
hardware registers and the stack designated for this purpose
by the Alpha calling standard.
A simple example of Macro32 argument passing follows:
.psect code,nowrt,exe,quad
$ssdef
.entry main,^m<r2>
pushl #^x10101
pushl #^x20202
pushl #^x30303
pushl #^x40404
pushl #^x50505
pushl #^x60606
pushl #^x70707
calls #^x7,sub
cmpl r0,#^x10101
beql 10$
movl #ss$_badparam,r0
ret
10$: movl #ss$_normal,r0
ret
.entry sub,^m<r2>
movl 7*4(ap),r0
ret
.end main
|