| 1 | /* Copyright (C) 2000-2022 Free Software Foundation, Inc. |
|---|
| 2 | This file is part of the GNU C Library. |
|---|
| 3 | |
|---|
| 4 | The GNU C Library is free software; you can redistribute it and/or |
|---|
| 5 | modify it under the terms of the GNU Lesser General Public |
|---|
| 6 | License as published by the Free Software Foundation; either |
|---|
| 7 | version 2.1 of the License, or (at your option) any later version. |
|---|
| 8 | |
|---|
| 9 | The GNU C Library is distributed in the hope that it will be useful, |
|---|
| 10 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
|---|
| 11 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|---|
| 12 | Lesser General Public License for more details. |
|---|
| 13 | |
|---|
| 14 | You should have received a copy of the GNU Lesser General Public |
|---|
| 15 | License along with the GNU C Library. If not, see |
|---|
| 16 | <https://www.gnu.org/licenses/>. */ |
|---|
| 17 | |
|---|
| 18 | /* Return pointer to first occurrence of CH in STR. */ |
|---|
| 19 | |
|---|
| 20 | #include <sysdep.h> |
|---|
| 21 | |
|---|
| 22 | .arch ev6 |
|---|
| 23 | .set noreorder |
|---|
| 24 | .set noat |
|---|
| 25 | |
|---|
| 26 | ENTRY(__rawmemchr) |
|---|
| 27 | #ifdef PROF |
|---|
| 28 | ldgp gp, 0(pv) |
|---|
| 29 | lda AT, _mcount |
|---|
| 30 | jsr AT, (AT), _mcount |
|---|
| 31 | .prologue 1 |
|---|
| 32 | #else |
|---|
| 33 | .prologue 0 |
|---|
| 34 | #endif |
|---|
| 35 | |
|---|
| 36 | ldq_u t0, 0(a0) # L : load first quadword Latency=3 |
|---|
| 37 | and a1, 0xff, t3 # E : 00000000000000ch |
|---|
| 38 | insbl a1, 1, t5 # U : 000000000000ch00 |
|---|
| 39 | insbl a1, 7, a2 # U : ch00000000000000 |
|---|
| 40 | |
|---|
| 41 | insbl t3, 6, a3 # U : 00ch000000000000 |
|---|
| 42 | or t5, t3, a1 # E : 000000000000chch |
|---|
| 43 | andnot a0, 7, v0 # E : align our loop pointer |
|---|
| 44 | lda t4, -1 # E : build garbage mask |
|---|
| 45 | |
|---|
| 46 | mskqh t4, a0, t4 # U : only want relevant part of first quad |
|---|
| 47 | or a2, a3, a2 # E : chch000000000000 |
|---|
| 48 | inswl a1, 2, t5 # E : 00000000chch0000 |
|---|
| 49 | inswl a1, 4, a3 # E : 0000chch00000000 |
|---|
| 50 | |
|---|
| 51 | or a1, a2, a1 # E : chch00000000chch |
|---|
| 52 | or a3, t5, t5 # E : 0000chchchch0000 |
|---|
| 53 | cmpbge zero, t4, t4 # E : bits set iff byte is garbage |
|---|
| 54 | nop # E : |
|---|
| 55 | |
|---|
| 56 | /* This quad is _very_ serialized. Lots of stalling happens */ |
|---|
| 57 | or t5, a1, a1 # E : chchchchchchchch |
|---|
| 58 | xor t0, a1, t1 # E : make bytes == c zero |
|---|
| 59 | cmpbge zero, t1, t0 # E : bits set iff byte == c |
|---|
| 60 | andnot t0, t4, t0 # E : clear garbage bits |
|---|
| 61 | |
|---|
| 62 | cttz t0, a2 # U0 : speculative (in case we get a match) |
|---|
| 63 | nop # E : |
|---|
| 64 | nop # E : |
|---|
| 65 | bne t0, $found # U : |
|---|
| 66 | |
|---|
| 67 | /* |
|---|
| 68 | * Yuk. This loop is going to stall like crazy waiting for the |
|---|
| 69 | * data to be loaded. Not much can be done about it unless it's |
|---|
| 70 | * unrolled multiple times, which is generally unsafe. |
|---|
| 71 | */ |
|---|
| 72 | $loop: |
|---|
| 73 | ldq t0, 8(v0) # L : Latency=3 |
|---|
| 74 | addq v0, 8, v0 # E : |
|---|
| 75 | xor t0, a1, t1 # E : |
|---|
| 76 | cmpbge zero, t1, t0 # E : bits set iff byte == c |
|---|
| 77 | |
|---|
| 78 | cttz t0, a2 # U0 : speculative (in case we get a match) |
|---|
| 79 | nop # E : |
|---|
| 80 | nop # E : |
|---|
| 81 | beq t0, $loop # U : |
|---|
| 82 | |
|---|
| 83 | $found: |
|---|
| 84 | negq t0, t1 # E : clear all but least set bit |
|---|
| 85 | and t0, t1, t0 # E : |
|---|
| 86 | addq v0, a2, v0 # E : Add in the bit number from above |
|---|
| 87 | ret # L0 : |
|---|
| 88 | |
|---|
| 89 | END(__rawmemchr) |
|---|
| 90 | |
|---|
| 91 | libc_hidden_def (__rawmemchr) |
|---|
| 92 | weak_alias (__rawmemchr, rawmemchr) |
|---|
| 93 | |
|---|
