| t@@ -23,32 +23,27 @@
*/
.text
-/* XXX: had to use "-4(%esp)" kludge to get around inability to
- * push/pop without first adjusting %esp. This may not be
- * as fast as using push/pop (and accessing pushed element
- * with "(%esp)".)
- */
-
.p2align 2,0x90
.globl mpvecdigmulsub
.type mpvecdigmulsub, @function
mpvecdigmulsub:
/* Prelude */
- pushl %ebp
- movl %ebx, -8(%esp) /* save on stack */
- movl %esi, -12(%esp)
- movl %edi, -16(%esp)
+ pushl %ebp /* save on stack */
+ pushl %ebx
+ pushl %esi
+ pushl %edi
- movl 8(%esp), %esi /* b */
- movl 12(%esp), %ecx /* n */
- movl 16(%esp), %ebx /* m */
- movl 20(%esp), %edi /* p */
+ leal 20(%esp), %ebp /* %ebp = FP for now */
+ movl 0(%ebp), %esi /* b */
+ movl 4(%ebp), %ecx /* n */
+ movl 8(%ebp), %ebx /* m */
+ movl 12(%ebp), %edi /* p */
xorl %ebp, %ebp
- movl %ebp, -4(%esp)
+ pushl %ebp
_mulsubloop:
movl (%esi, %ebp, 4),%eax /* lo = b[i] */
mull %ebx /* hi, lo = b[i] * m */
- addl -4(%esp), %eax /* lo += oldhi */
+ addl 0(%esp), %eax /* lo += oldhi */
jae _mulsubnocarry1
incl %edx /* hi += carry */
_mulsubnocarry1:
t@@ -56,23 +51,21 @@ _mulsubnocarry1:
jae _mulsubnocarry2
incl %edx /* hi += carry */
_mulsubnocarry2:
- movl %edx, -4(%esp)
+ movl %edx, 0(%esp)
incl %ebp
loop _mulsubloop
- movl -4(%esp), %eax
+ popl %eax
subl %eax, (%edi, %ebp, 4)
jae _mulsubnocarry3
movl $-1, %eax
jmp done
-
_mulsubnocarry3:
movl $1, %eax
-
done:
/* Postlude */
- movl -8(%esp), %ebx /* restore from stack */
- movl -12(%esp), %esi
- movl -16(%esp), %edi
- movl %esp, %ebp
- leave
+ popl %edi
+ popl %esi
+ popl %ebx
+ popl %ebp
ret
+ |