openbios/openboot/obp/fm/kernel/sparc/moveslow.fth
Branch
-
Show log
Commit
-
Hash : 1a08f436
Author :
Date : 2010-08-25T09:23:17
change to well-formed svn directory structure. git-svn-id: svn://coreboot.org/openboot/trunk@2 4486e004-3823-0410-90c7-fb508cc143bb
-
obp/fm/kernel/sparc/moveslow.fth
-
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
\ moveslow.fth 1.5 93/10/21 \ Copyright 1985-1990 Bradley Forthware \ Mike Saari's blazing `move' ... \ This implements the MOVE instruction. It is optimized \ for speed, particularly when longword stores may be used. \ This version omits the doubleword optimization for machines where \ doubleword operations do not work. code (move) ( src dst cnt -- ) \ tos = Count sp 1 /n* scr nget \ scr = Src address sp 0 /n* sc1 nget \ sc1 = Dst address \ sc2 = Temp. data being transferred \ sc3 = src xor drc, low bits=0 indicates compatible \ sc4 = Working src in loops \ (also temp last+1 address) \ sc5 = Working dst in loops \ sc6 = Loop index scr sc1 %g0 subcc \ Src > dst? > if \ Then copy low-to-high scr sc1 sc3 xor \ (delay) sc3 low bits=0 indicates compatible tos h# 10 %g0 subcc \ Enough bytes to bother optimizing? >= if \ Otherwise, just skip to byte move sc3 1 %g0 andcc \ (delay) =0 if at least shortword aligned 0= if \ Otherwise, just skip to byte move scr 1 %g0 andcc \ (delay) Not on halfword boundary? 0<> if \ Ensure halfword alignment (lower) scr 0 sc2 ldub \ (delay) Load bottom byte sc2 sc1 0 stb \ Store byte scr 1 scr add \ Advance by one byte sc1 1 sc1 add \ " tos 1 tos sub \ Decrement count then sc3 2 %g0 andcc \ =0 if at least longword aligned 0= if \ Otherwise, skip to halfword case scr 2 %g0 andcc \ (delay) Not on longword boundary? 0<> if \ Ensure longword alignment (lower) scr 0 sc2 lduh \ (delay) Load bottom halfword sc2 sc1 0 sth \ Store halfword scr 2 scr add \ Advance by one halfword sc1 2 sc1 add \ " tos 2 tos sub \ Decrement count then \ Longword Copy Loop (low-to-high) tos 3 sc6 andn \ Index w/ even multiples of 4 scr sc6 scr add \ src = src+index scr 4 sc4 sub \ Working src = src+index-4 sc1 sc6 sc1 add \ dst = dst+index sc1 4 sc5 sub \ Working dst = dst+index-4 %g0 sc6 sc6 subcc \ Negate index begin < while sc6 4 sc6 addcc \ (delay) Increment index sc4 sc6 sc2 ld \ Load longword repeat sc2 sc5 sc6 st \ (delay) Store longword tos 3 tos and \ At end, adjust cnt for few remaining else \ Halfword Copy Loop (low-to-high) nop \ (delay) tos 1 sc6 andn \ Index w/ even multiples of 2 scr sc6 scr add \ src = src+index scr 2 sc4 sub \ Working src = src+index-2 sc1 sc6 sc1 add \ dst = dst+index sc1 2 sc5 sub \ Working dst = dst+index-2 %g0 sc6 sc6 subcc \ Negate index begin < while sc6 2 sc6 addcc \ (delay) Increment index sc4 sc6 sc2 lduh \ Load halfword repeat sc2 sc5 sc6 sth \ (delay) Store halfword tos 1 tos and \ At end, adjust cnt for few remaining then then then \ Now do a normal byte move for all remaining bytes (at top) \ Byte Copy Loop (low-to-high) \ (tos = index) scr tos scr add scr 1 sc4 sub \ Working src = src+cnt-1 sc1 tos sc1 add sc1 1 sc5 sub \ Working dst = dst+cnt-1 %g0 tos sc6 subcc \ Negate index begin < while sc6 1 sc6 addcc \ (delay) Increment cnt sc4 sc6 sc2 ldub \ Load byte repeat sc2 sc5 sc6 stb \ (delay) Store byte else \ Copy high-to-low case nop \ (delay) tos h# 10 %g0 subcc \ Enough bytes to bother optimizing? >= if \ Otherwise, just skip to byte move sc3 1 %g0 andcc \ (delay) =0 if at least shortword aligned 0= if \ Otherwise, just skip to byte move scr tos sc4 add \ (delay) Calculate last+1 address sc4 1 %g0 andcc \ Not on halfword boundary? (at top) 0<> if \ Ensure halfword alignment (at top) sc4 -1 sc2 ldub \ (delay) Load top byte tos 1 tos sub \ Decrement count sc2 sc1 tos stb \ Store byte sc4 1 sc4 sub \ Recalculate last+1 address then sc3 2 %g0 andcc \ =0 if at least longword aligned 0= if \ Otherwise, skip to halfword case sc4 2 %g0 andcc \ (delay) Not on longword boundary? (at top) 0<> if \ Ensure longword alignment (at top) sc4 -2 sc2 lduh \ (delay) Load top halfword tos 2 tos sub \ Decrement count sc2 sc1 tos sth \ Store halfword sc4 2 sc4 sub \ Recalculate last+1 address then \ Longword Copy Loop (high-to-low) scr 4 sc4 add \ Working src = src+4 sc1 4 sc5 add \ Working dst = dst+4 tos 4 sc6 subcc \ Loop index = cnt-4 begin >= while sc6 4 sc6 subcc \ (delay) Decrement index sc4 sc6 sc2 ld \ Load longword repeat sc2 sc5 sc6 st \ (delay) Store longword tos 3 tos and \ At end, adjust cnt for few remaining else \ Halfword Copy Loop (high-to-low) nop \ (delay) scr 2 sc4 add \ Working src = src+2 sc1 2 sc5 add \ Working dst = dst+2 tos 2 sc6 subcc \ Loop index = cnt-2 begin >= while sc6 2 sc6 subcc \ (delay) Decrement index sc4 sc6 sc2 lduh \ Load halfword repeat sc2 sc5 sc6 sth \ (delay) Store halfword tos 1 tos and \ At end, adjust cnt for few remaining then then then \ Now do a normal byte move for all remaining bytes (at bottom) \ Byte Copy Loop (high-to-low) scr 1 sc4 add \ Working src = src+1 sc1 1 sc5 add \ Working dst = dst+1 tos 1 tos subcc \ Loop index = cnt-1 begin >= while tos 1 tos subcc \ (delay) Decrement index sc4 tos sc2 ldub \ Load byte repeat sc2 sc5 tos stb \ (delay) Store byte then sp 2 /n* tos nget \ Delete 3 stack items sp 3 /n* sp add \ " c; defer move ' (move) is move