projects / thirdparty / glibc.git / blame
| Commit | Line | Data |
|---|---|---|
| f345d77f UD |
1 | /* Copyright (C) 2000 Free Software Foundation, Inc. |
| 2 | ||
| 3 | This file is part of the GNU C Library. | |
| 4 | ||
| 5 | The GNU C Library is free software; you can redistribute it and/or | |
| 6 | modify it under the terms of the GNU Library General Public License as | |
| 7 | published by the Free Software Foundation; either version 2 of the | |
| 8 | License, or (at your option) any later version. | |
| 9 | ||
| 10 | The GNU C Library is distributed in the hope that it will be useful, | |
| 11 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
| 13 | Library General Public License for more details. | |
| 14 | ||
| 15 | You should have received a copy of the GNU Library General Public | |
| 16 | License along with the GNU C Library; see the file COPYING.LIB. If not, | |
| 17 | write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
| 18 | Boston, MA 02111-1307, USA. */ | |
| 19 | ||
| 20 | /* Return pointer to first occurrence of CH in STR. */ | |
| 21 | ||
| 22 | #include <sysdep.h> | |
| 23 | ||
| 24 | .arch ev6 | |
| 25 | .set noreorder | |
| 26 | .set noat | |
| 27 | ||
| 28 | ENTRY(__rawmemchr) | |
| 29 | #ifdef PROF | |
| 30 | ldgp gp, 0(pv) | |
| 31 | lda AT, _mcount | |
| 32 | jsr AT, (AT), _mcount | |
| 33 | .prologue 1 | |
| 34 | #else | |
| 35 | .prologue 0 | |
| 36 | #endif | |
| 37 | ||
| 38 | ldq_u t0, 0(a0) # L : load first quadword Latency=3 | |
| 39 | and a1, 0xff, t3 # E : 00000000000000ch | |
| 40 | insbl a1, 1, t5 # U : 000000000000ch00 | |
| 41 | insbl a1, 7, a2 # U : ch00000000000000 | |
| 42 | ||
| 43 | insbl t3, 6, a3 # U : 00ch000000000000 | |
| 44 | or t5, t3, a1 # E : 000000000000chch | |
| 45 | andnot a0, 7, v0 # E : align our loop pointer | |
| 46 | lda t4, -1 # E : build garbage mask | |
| 47 | ||
| 48 | mskqh t4, a0, t4 # U : only want relevant part of first quad | |
| 49 | or a2, a3, a2 # E : chch000000000000 | |
| 50 | inswl a1, 2, t5 # E : 00000000chch0000 | |
| 51 | inswl a1, 4, a3 # E : 0000chch00000000 | |
| 52 | ||
| 53 | or a1, a2, a1 # E : chch00000000chch | |
| 54 | or a3, t5, t5 # E : 0000chchchch0000 | |
| 55 | cmpbge zero, t4, t4 # E : bits set iff byte is garbage | |
| 56 | nop # E : | |
| 57 | ||
| 58 | /* This quad is _very_ serialized. Lots of stalling happens */ | |
| 59 | or t5, a1, a1 # E : chchchchchchchch | |
| 60 | xor t0, a1, t1 # E : make bytes == c zero | |
| 61 | cmpbge zero, t1, t0 # E : bits set iff byte == c | |
| 62 | andnot t0, t4, t0 # E : clear garbage bits | |
| 63 | ||
| 64 | cttz t0, a2 # U0 : speculative (in case we get a match) | |
| 65 | nop # E : | |
| 66 | nop # E : | |
| 67 | bne t0, $found # U : | |
| 68 | ||
| 69 | /* | |
| 70 | * Yuk. This loop is going to stall like crazy waiting for the | |
| 71 | * data to be loaded. Not much can be done about it unless it's | |
| 72 | * unrolled multiple times, which is generally unsafe. | |
| 73 | */ | |
| 74 | $loop: | |
| 75 | ldq t0, 8(v0) # L : Latency=3 | |
| 76 | addq v0, 8, v0 # E : | |
| 77 | xor t0, a1, t1 # E : | |
| 78 | cmpbge zero, t1, t0 # E : bits set iff byte == c | |
| 79 | ||
| 80 | cttz t0, a2 # U0 : speculative (in case we get a match) | |
| 81 | nop # E : | |
| 82 | nop # E : | |
| 83 | beq t0, $loop # U : | |
| 84 | ||
| 85 | $found: | |
| 86 | negq t0, t1 # E : clear all but least set bit | |
| 87 | and t0, t1, t0 # E : | |
| 88 | addq v0, a2, v0 # E : Add in the bit number from above | |
| 89 | ret # L0 : | |
| 90 | ||
| 91 | END(__rawmemchr) | |
| 92 | ||
| 93 | weak_alias (__rawmemchr, rawmemchr) |