encoding/json: remove a bounds check in readValue

readValue is a hot function, clocking in at ~13% flat CPU use in
CodeDecoder. In particular, looping over the bytes is slow. That's
partially because the code contains a bounds check at the start of the
loop.

The source of the problem is that scanp is a signed integer, and comes
from a field, so the compiler doesn't know that it's non-negative. Help
it with a simple and comparatively cheap hint.

While at it, use scanp as the index variable directly, removing the need
for a duplicate index variable which is later added back into scanp.

name           old time/op    new time/op    delta
CodeDecoder-8    11.3ms ± 1%    11.2ms ± 1%  -0.98%  (p=0.000 n=9+9)

name           old speed      new speed      delta
CodeDecoder-8   172MB/s ± 1%   174MB/s ± 1%  +0.99%  (p=0.000 n=9+9)

Updates #28923.

Change-Id: I138f83babdf316fc97697cc18f595c3403c1ddb7
Reviewed-on: https://go-review.googlesource.com/c/go/+/170939
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

diff --git a/src/encoding/json/stream.go b/src/encoding/json/stream.go
index 7d5137fbc716f289f273eac02bd7b39e44f332f7..e29127499b638a1328cae467ccac866ec618088f 100644 (file)
--- a/src/encoding/json/stream.go
+++ b/src/encoding/json/stream.go
@@ -92,20 +92,23 @@ func (dec *Decoder) readValue() (int, error) {
        scanp := dec.scanp
        var err error
 Input:
-       for {
+       // help the compiler see that scanp is never negative, so it can remove
+       // some bounds checks below.
+       for scanp >= 0 {
+
                // Look in the buffer for a new value.
-               for i, c := range dec.buf[scanp:] {
+               for ; scanp < len(dec.buf); scanp++ {
+                       c := dec.buf[scanp]
                        dec.scan.bytes++
                        switch dec.scan.step(&dec.scan, c) {
                        case scanEnd:
-                               scanp += i
                                break Input
                        case scanEndObject, scanEndArray:
                                // scanEnd is delayed one byte.
                                // We might block trying to get that byte from src,
                                // so instead invent a space byte.
                                if stateEndValue(&dec.scan, ' ') == scanEnd {
-                                       scanp += i + 1
+                                       scanp++
                                        break Input
                                }
                        case scanError:
@@ -113,7 +116,6 @@ Input:
                                return 0, dec.scan.err
                        }
                }
-               scanp = len(dec.buf)
 
                // Did the last read have an error?
                // Delayed until now to allow buffer scan.