d0p1/ack
1
0
Fork 0
Code Issues Pull requests Packages Projects Releases Wiki Activity
ack/mach/m68020/libem/divrem8.s

77 lines
1.8 KiB
ArmAsm
Raw Permalink Normal View History

Add 8-byte long long for linux68k. Add rules for 8-byte integers to m68020 ncg. Add 8-byte long long to ACK C on linux68k. Enable long-long tests for linux68k. The tests pass in our emulator using musahi; I don't have a real 68k processor and haven't tried other emulators. Still missing are conversions between 8-byte integers and any size of floats. The long-long tests don't cover these conversions, and our emulator can't do floating-point. Our build always enables TBL68020 and uses word size 4. Without TBL68020, 8-byte multiply and divide are missing. With word size 2, some conversions between 2-byte and 8-byte integers are missing. Fix .cii in libem, which didn't work when converting from 1-byte or 2-byte integers. Now .cii and .cuu work, but also add some rules to skip .cii and .cuu when converting 8-byte integers. The new rule for loc 4 loc 8 cii `with test_set4` exposes a bug: the table may believe that the condition codes test a 4-byte register when they only test a word or byte, and this incorrect test may describe an unsigned word or byte as negative. Another rule `with exact test_set1+test_set2` works around the bug by ignoring the negative flag, because a zero-extended word or byte is never negative. The old rules for comparison and logic do work with 8-byte integers and bitsets, but add some specific 8-byte rules to skip libem calls or loops. There were no rules for 8-byte arithmetic, shift, or rotate; so add some. There is a register shortage, because the table requires preserving d3 to d7, leaving only 3 data registers (d0, d1, d2) for 8-byte operations. Because of the shortage, the code may move data to an address register, or read a memory location more than once. The multiplication and division code are translations of the i386 code. They pass the tests, but might not give the best performance on a real 68k processor.
2019-09-24 17:32:17 +00:00
.define .divrem8
.sect .text
.sect .rom
.sect .data
.sect .bss
yh=16
yl=20
xh=24
xl=28
! This private sub for .dvi8, .dvu8, .rmi8, .rmu8
! does unsigned division of x = xh:xl by y = yh:yl,
! yields d0:d1 = quotient, d2:d3 = remainder.
.sect .text
.divrem8:
! Caller must set d0, d1 like so:
! mov.l (xh, sp), d0
! mov.l (yh, sp), d1
tst.l d1
bne 1f ! branch if y >= 2**32
! y = yl, so x / y = xh:xl / yl = qh:0 + rh:xl / yl
! where qh, rh are quotient, remainder from xh / yl.
move.l (xl, sp), d1
move.l (yl, sp), d2
clr.l d3 ! d3:d0 = xh
divu.l d2, d3:d0 ! d0 = 0:xh / yl, d3 = rh
divu.l d2, d3:d1 ! d1 = rh:xl / yl, so d0:d1 = x / y
clr.l d2 ! remainder in d2:d3
rts
1: ! Here y >= 2**32.
move.l d0, a0 ! save xh
move.l d1, a1 ! save yh
move.l d7, a2 ! save caller's d7
! Find y >> right in [2**31, 2**32).
move.l (yl, sp), d2
bfffo d1[0:32], d3 ! find highest set bit in yh
lsl.l d3, d1 ! shift yh left
bset #5, d3
neg.l d3 ! right = (32 - left) modulo 64
lsr.l d3, d2 ! shift yl right
or.l d1, d2 ! d2 = y >> right
! Estimate x / y as q = (x / (y >> right)) >> right.
move.l (xl, sp), d1
clr.l d7
divu.l d2, d7:d0
divu.l d2, d7:d1 ! d0:d1 = x / (y >> right)
lsr.l d3, d1
bset #5, d3
neg.l d3
lsl.l d3, d0
or.l d0, d1 ! d1 = q
! Calculate the remainder x - y * q. If the subtraction
! overflows, then the correct quotient is q - 1, else it is q.
move.l a1, d3 ! yh
mulu.l d1, d3 ! yh * q
move.l (yl, sp), d7
mulu.l d1, d0:d7 ! yl * q
add.l d3, d0 ! d0:d7 = y * q
move.l (xl, sp), d3
move.l a0, d2 ! d2:d3 = x
sub.l d7, d3
subx.l d0, d2 ! d2:d3 = x - y * q
bcc 1f ! branch unless subtraction overflowed
sub.l #1, d1 ! fix quotient
move.l a1, d7 ! yh
add.l (yl, sp), d3
addx.l d7, d2 ! fix remainder
1: clr.l d0 ! d0:d1 = quotient
move.l a2, d7 ! restore caller's d7
rts
Reference in a new issue Copy permalink