diff options
| author | Hang Su <darcy.sh@antfin.com> | 2019-09-18 13:35:23 +0800 |
|---|---|---|
| committer | Hang Su <darcy.sh@antfin.com> | 2019-09-19 22:16:52 +0800 |
| commit | d72c63664b46af52f4880a2a015666a52deb0749 (patch) | |
| tree | c482e7a2686e4e6287ea4e8391d5f31affbaba5e /pkg/sentry/usermem | |
| parent | 28f431335b182519f420e026edac8b7bcfd2a40a (diff) | |
Accelerate byte lookup in string with `bytealg/indexbyte`
`bytealg/indexbyte` will use AVX or SSE instruction set, if possible,
which could accelerate `CopyStringIn` function by 28%.
In worst case(CPU doesn't support SSE), `bytealg/indexbyte`
will degenerate to traversal lookup. When dealing with
short strings, `bytealg/indexbyte` has the same performance level as
before.
Signed-off-by: Jianfeng Tan <henry.tjf@antfin.com>
Signed-off-by: Hang Su <darcy.sh@antfin.com>
Diffstat (limited to 'pkg/sentry/usermem')
| -rw-r--r-- | pkg/sentry/usermem/usermem.go | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/pkg/sentry/usermem/usermem.go b/pkg/sentry/usermem/usermem.go index 6eced660a..7b1f312b1 100644 --- a/pkg/sentry/usermem/usermem.go +++ b/pkg/sentry/usermem/usermem.go @@ -16,6 +16,7 @@ package usermem import ( + "bytes" "errors" "io" "strconv" @@ -270,11 +271,10 @@ func CopyStringIn(ctx context.Context, uio IO, addr Addr, maxlen int, opts IOOpt n, err := uio.CopyIn(ctx, addr, buf[done:done+readlen], opts) // Look for the terminating zero byte, which may have occurred before // hitting err. - for i, c := range buf[done : done+n] { - if c == 0 { - return stringFromImmutableBytes(buf[:done+i]), nil - } + if i := bytes.IndexByte(buf[done:done+n], byte(0)); i >= 0 { + return stringFromImmutableBytes(buf[:done+i]), nil } + done += n if err != nil { return stringFromImmutableBytes(buf[:done]), err |