diff options
Diffstat (limited to 'volk/spu_lib/spu_memset_unaligned.S')
| -rw-r--r-- | volk/spu_lib/spu_memset_unaligned.S | 44 |
1 files changed, 22 insertions, 22 deletions
diff --git a/volk/spu_lib/spu_memset_unaligned.S b/volk/spu_lib/spu_memset_unaligned.S index a655c4c52..c260a125c 100644 --- a/volk/spu_lib/spu_memset_unaligned.S +++ b/volk/spu_lib/spu_memset_unaligned.S @@ -1,19 +1,19 @@ /* -*- asm -*- */ /* * Copyright 2008 Free Software Foundation, Inc. - * + * * This file is part of GNU Radio - * + * * GNU Radio is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 3, or (at your option) * any later version. - * + * * GNU Radio is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. - * + * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. @@ -33,11 +33,11 @@ * size_t i; * for (i = 0; i < n; i++) * p[i] = c; - * + * * return pv; * } */ - + #define p_arg arg1 // we're going to clobber arg1 w/ the return value #define c arg2 // the constant we're writing #define n arg3 // how many bytes to write @@ -50,16 +50,16 @@ #define an r18 // aligned n (n rounded down to mod 16 boundary) #define next_p r19 #define cond1 r20 -#define cond2 r21 +#define cond2 r21 #define m r22 #define r r23 - + PROC_ENTRY(libvector_memset_unaligned) - + // Hint the return from do_head, in case we go that way. // There's pretty much nothing to can do to hint the branch to it. hbrr do_head_br, head_complete - + MR(p, p_arg) // leaves p, the return value, in the correct reg (r3) BRZ_RETURN(n) @@ -69,11 +69,11 @@ head_complete: /* - * preconditions: + * preconditions: * p%16 == 0, n > 0 */ hbrr middle_loop_br, middle_loop - + ROUND_DOWN(an, n, 16) // an is "aligned n" MODULO(n, n, 16) // what's left over in the last quad brz an, do_tail // no whole quad words; skip to tail @@ -96,20 +96,20 @@ middle_loop: stqd c, 4*16(p) stqd c, 5*16(p) stqd c, 6*16(p) - + MR(p, next_p) stqd c, 7*16-128(next_p) or cond2, n, an middle_loop_br: brnz cond1, middle_loop - + /* - * if an and n are both zero, return now + * if an and n are both zero, return now */ BRZ_RETURN(cond2) /* - * otherwise handle last of full quad words + * otherwise handle last of full quad words * * 0 <= an < 128, p%16 == 0 */ @@ -119,18 +119,18 @@ middle2: */ brz an, do_tail hbrr middle2_loop_br, middle2_loop - + .p2align 3 -middle2_loop: +middle2_loop: ai next_p, p, 16 stqd c, 0(p) ai an, an, -16 LMR(p, next_p) middle2_loop_br: brnz an, middle2_loop - + /* We're done with the full quadwords. */ - + /* * Handle the final partial quadword. * We'll be modifying only the left hand portion of the quad. @@ -146,7 +146,7 @@ do_tail: shlqby mask, mask, t1 selb t0, old, c, mask stqd t0, 0(p) -do_tail_ret: +do_tail_ret: RETURN() /* @@ -176,7 +176,7 @@ do_head: MR(t1, p) sf t0, m, r // t0 = r - m a p, p, m // p += m - rotqby mask, mask, t0 // rotate 0's to the right place + rotqby mask, mask, t0 // rotate 0's to the right place sf n, m, n // n -= m selb t0, c, old, mask // merge stqd t0, 0(t1) |