Lines Matching +full:16 +full:- +full:17
1 /* SPDX-License-Identifier: GPL-2.0 */
3 * arch/alpha/lib/ev6-memset.S
8 * 21264 version contributed by Rick Gorton <rick.gorton@alpha-processor.com>
13 * ftp.digital.com/pub/Digital/info/semiconductor/literature/dsc-library.html
15 * E - either cluster
16 * U - upper subcluster; U0 - subcluster U0; U1 - subcluster U1
17 * L - lower subcluster; L0 - subcluster L0; L1 - subcluster L1
48 * undertake a major re-write to interleave the constant materialization
49 * with other parts of the fall-through code. This is important, even
53 and $17,255,$1 # E : 00000000000000ch
54 insbl $17,1,$2 # U : 000000000000ch00
55 bis $16,$16,$0 # E : return value
58 addq $18,$16,$6 # E : max address to write to
59 bis $1,$2,$17 # E : 000000000000chch
64 inswl $17,4,$5 # U : 0000chch00000000
65 xor $16,$6,$1 # E : will complete write be within one quadword?
66 inswl $17,6,$2 # U : chch000000000000
68 or $17,$3,$17 # E : 00000000chchchch
71 and $16,7,$3 # E : Target addr misalignment
73 or $17,$2,$17 # E : chchchchchchchch
81 ldq_u $4,0($16) # L : Fetch first partial
82 bis $16,$16,$5 # E : Save the address
83 insql $17,$16,$2 # U : Insert new bytes
87 mskql $4,$16,$4 # U : clear relevant parts of the quad
88 subq $16,$3,$16 # E : $16 is new aligned destination
105 bis $16,$16,$5 # E : Save dest address
112 * $16 Current destination address
113 * $5 A copy of $16
119 and $16, 0x3f, $2 # E : Forward work (only useful for unrolled loop)
120 subq $3, 16, $4 # E : Only try to unroll if > 128 bytes
125 * We know we've got at least 16 quads, minimum of one trip
136 stq $17, 0($5) # L :
148 * $3 - number quads left to go
149 * $5 - target address (aligned 0mod64)
150 * $17 - mask of stuff to store
163 stq $17, 0($5) # L :
167 stq $17, 8($5) # L :
168 stq $17, 16($5) # L :
171 stq $17, 24($5) # L :
172 stq $17, 32($5) # L :
176 stq $17, 40($5) # L :
177 stq $17, 48($5) # L :
178 subq $3, 16, $2 # E : Repeat the loop at least once more?
181 stq $17, 56($5) # L :
197 stq $17,0($5) # L :
211 insqh $17,$6,$4 # U : New bits
217 ldq_u $1,0($16) # L :
218 insql $17,$16,$2 # U : New bits
219 mskql $1,$16,$4 # U : Clear old
225 stq_u $1,0($16) # L :
246 addq $18,$16,$6 # E : max address to write to
247 bis $16,$16,$0 # E : return value
248 xor $16,$6,$1 # E : will complete write be within one quadword?
253 and $16,7,$3 # E : Target addr misalignment
259 ldq_u $4,0($16) # L : Fetch first partial
260 bis $16,$16,$5 # E : Save the address
261 insql $17,$16,$2 # U : Insert new bytes
265 mskql $4,$16,$4 # U : clear relevant parts of the quad
266 subq $16,$3,$16 # E : $16 is new aligned destination
283 bis $16,$16,$5 # E : Save dest address
290 * $16 Current destination address
291 * $5 A copy of $16
297 and $16, 0x3f, $2 # E : Forward work (only useful for unrolled loop)
298 subq $3, 16, $4 # E : Only try to unroll if > 128 bytes
303 * We know we've got at least 16 quads, minimum of one trip
314 stq $17, 0($5) # L :
326 * $3 - number quads left to go
327 * $5 - target address (aligned 0mod64)
328 * $17 - mask of stuff to store
341 stq $17, 0($5) # L :
345 stq $17, 8($5) # L :
346 stq $17, 16($5) # L :
349 stq $17, 24($5) # L :
350 stq $17, 32($5) # L :
354 stq $17, 40($5) # L :
355 stq $17, 48($5) # L :
356 subq $3, 16, $2 # E : Repeat the loop at least once more?
359 stq $17, 56($5) # L :
375 stq $17,0($5) # L :
389 insqh $17,$6,$4 # U : New bits
395 ldq_u $1,0($16) # L :
396 insql $17,$16,$2 # U : New bits
397 mskql $1,$16,$4 # U : Clear old
403 stq_u $1,0($16) # L :
424 inswl $17,0,$5 # U : 000000000000c1c2
425 inswl $17,2,$2 # U : 00000000c1c20000
426 bis $16,$16,$0 # E : return value
427 addq $18,$16,$6 # E : max address to write to
430 inswl $17,4,$3 # U : 0000c1c200000000
431 inswl $17,6,$4 # U : c1c2000000000000
432 xor $16,$6,$1 # E : will complete write be within one quadword?
435 or $3,$4,$17 # E : c1c2c1c200000000
437 and $16,7,$3 # E : Target addr misalignment
439 or $17,$2,$17 # E : c1c2c1c2c1c2c1c2
447 ldq_u $4,0($16) # L : Fetch first partial
448 bis $16,$16,$5 # E : Save the address
449 insql $17,$16,$2 # U : Insert new bytes
453 mskql $4,$16,$4 # U : clear relevant parts of the quad
454 subq $16,$3,$16 # E : $16 is new aligned destination
471 bis $16,$16,$5 # E : Save dest address
478 * $16 Current destination address
479 * $5 A copy of $16
485 and $16, 0x3f, $2 # E : Forward work (only useful for unrolled loop)
486 subq $3, 16, $4 # E : Only try to unroll if > 128 bytes
491 * We know we've got at least 16 quads, minimum of one trip
502 stq $17, 0($5) # L :
514 * $3 - number quads left to go
515 * $5 - target address (aligned 0mod64)
516 * $17 - mask of stuff to store
529 stq $17, 0($5) # L :
533 stq $17, 8($5) # L :
534 stq $17, 16($5) # L :
537 stq $17, 24($5) # L :
538 stq $17, 32($5) # L :
542 stq $17, 40($5) # L :
543 stq $17, 48($5) # L :
544 subq $3, 16, $2 # E : Repeat the loop at least once more?
547 stq $17, 56($5) # L :
563 stq $17,0($5) # L :
577 insqh $17,$6,$4 # U : New bits
583 ldq_u $1,0($16) # L :
584 insql $17,$16,$2 # U : New bits
585 mskql $1,$16,$4 # U : Clear old
591 stq_u $1,0($16) # L :