internal/bytealg: share code for IndexByte functions on arm

Move the shared code of IndexByte and IndexByteString into
indexbytebody. This will allow to implement optimizations (e.g.
for #29001) in a single function.

Change-Id: I1d550da8eb65f95e492a460a12058cc35b1162b6
Reviewed-on: https://go-review.googlesource.com/c/go/+/167939
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

diff --git a/src/internal/bytealg/indexbyte_arm.s b/src/internal/bytealg/indexbyte_arm.s
index 7d9bbb183d803c30a2e86cc39a169b0077c2ff33..faf97977a63d92a3490d13b4143a62fde19c4629 100644 (file)
--- a/src/internal/bytealg/indexbyte_arm.s
+++ b/src/internal/bytealg/indexbyte_arm.s
@@ -9,46 +9,38 @@ TEXT ·IndexByte(SB),NOSPLIT,$0-20
        MOVW    b_base+0(FP), R0
        MOVW    b_len+4(FP), R1
        MOVBU   c+12(FP), R2    // byte to find
-       MOVW    R0, R4          // store base for later
-       ADD     R0, R1          // end
-
-_loop:
-       CMP     R0, R1
-       B.EQ    _notfound
-       MOVBU.P 1(R0), R3
-       CMP     R2, R3
-       B.NE    _loop
-
-       SUB     $1, R0          // R0 will be one beyond the position we want
-       SUB     R4, R0          // remove base
-       MOVW    R0, ret+16(FP)
-       RET
-
-_notfound:
-       MOVW    $-1, R0
-       MOVW    R0, ret+16(FP)
-       RET
+       MOVW    $ret+16(FP), R5
+       B       indexbytebody<>(SB)
 
 TEXT ·IndexByteString(SB),NOSPLIT,$0-16
        MOVW    s_base+0(FP), R0
        MOVW    s_len+4(FP), R1
        MOVBU   c+8(FP), R2     // byte to find
+       MOVW    $ret+12(FP), R5
+       B       indexbytebody<>(SB)
+
+// input:
+//  R0: data
+//  R1: data length
+//  R2: byte to find
+//  R5: address to put result
+TEXT indexbytebody<>(SB),NOSPLIT,$0-0
        MOVW    R0, R4          // store base for later
        ADD     R0, R1          // end
 
-_sib_loop:
+loop:
        CMP     R0, R1
-       B.EQ    _sib_notfound
+       B.EQ    notfound
        MOVBU.P 1(R0), R3
        CMP     R2, R3
-       B.NE    _sib_loop
+       B.NE    loop
 
        SUB     $1, R0          // R0 will be one beyond the position we want
        SUB     R4, R0          // remove base
-       MOVW    R0, ret+12(FP)
+       MOVW    R0, (R5)
        RET
 
-_sib_notfound:
+notfound:
        MOVW    $-1, R0
-       MOVW    R0, ret+12(FP)
+       MOVW    R0, (R5)
        RET