[dev.link] cmd/link/internal/loader: add 1-element object cache for use in toLocal

To speed up the loader.Loader.toLocal() method, cache the index of the
most recently accessed object file and check that object's sym range
in toLocal() before doing a full binary search over all object symbol
ranges. This speeds up relink of kubernetes/hyperkube by about 2%, and
improves compilebench (relative to the dev.link branch) by about 5%:

name                      old time/op       new time/op       delta
LinkCompiler                    1.62s ± 8%        1.50s ± 9%  -7.21%  (p=0.000 n=20+19)
LinkWithoutDebugCompiler        1.13s ± 8%        1.09s ±12%    ~     (p=0.052 n=20+20)

name                      old user-time/op  new user-time/op  delta
LinkCompiler                    1.94s ±18%        1.97s ±16%    ~     (p=0.813 n=19+20)
LinkWithoutDebugCompiler        1.15s ±16%        1.13s ±12%    ~     (p=0.547 n=20+20)

Change-Id: Id5a8a847b533858373c0462f03972d436eda6748
Reviewed-on: https://go-review.googlesource.com/c/go/+/204337
Reviewed-by: Jeremy Faller <jeremy@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

diff --git a/src/cmd/link/internal/loader/loader.go b/src/cmd/link/internal/loader/loader.go
index 52809c63da992f31d18eeda3caf92a8689f49d79..95e3005af2ec8f15a657e02f415b0d32539e0a09 100644 (file)
--- a/src/cmd/link/internal/loader/loader.go
+++ b/src/cmd/link/internal/loader/loader.go
@@ -65,6 +65,7 @@ type oReader struct {
 type objIdx struct {
        r *oReader
        i Sym // start index
+       e Sym // end index
 }
 
 type nameVer struct {
@@ -98,6 +99,7 @@ type Loader struct {
        extStart    Sym              // from this index on, the symbols are externally defined
        extSyms     []nameVer        // externally defined symbols
        builtinSyms []Sym            // global index of builtin symbols
+       ocache      int              // index (into 'objs') of most recent lookup
 
        symsByName    [2]map[string]Sym // map symbol name to index, two maps are for ABI0 and ABIInternal
        extStaticSyms map[nameVer]Sym   // externally defined static symbols, keyed by name
@@ -116,7 +118,7 @@ func NewLoader() *Loader {
        nbuiltin := goobj2.NBuiltin()
        return &Loader{
                start:         make(map[*oReader]Sym),
-               objs:          []objIdx{{nil, 0}},
+               objs:          []objIdx{{nil, 0, 0}},
                symsByName:    [2]map[string]Sym{make(map[string]Sym), make(map[string]Sym)},
                objByPkg:      make(map[string]*oReader),
                overwrite:     make(map[Sym]Sym),
@@ -143,7 +145,7 @@ func (l *Loader) addObj(pkg string, r *oReader) Sym {
        n := r.NSym() + r.NNonpkgdef()
        i := l.max + 1
        l.start[r] = i
-       l.objs = append(l.objs, objIdx{r, i})
+       l.objs = append(l.objs, objIdx{r, i, i + Sym(n) - 1})
        l.max += Sym(n)
        return i
 }
@@ -249,12 +251,17 @@ func (l *Loader) toLocal(i Sym) (*oReader, int) {
        if l.isExternal(i) {
                return nil, int(i - l.extStart)
        }
+       oc := l.ocache
+       if oc != 0 && i >= l.objs[oc].i && i <= l.objs[oc].e {
+               return l.objs[oc].r, int(i - l.objs[oc].i)
+       }
        // Search for the local object holding index i.
        // Below k is the first one that has its start index > i,
        // so k-1 is the one we want.
        k := sort.Search(len(l.objs), func(k int) bool {
                return l.objs[k].i > i
        })
+       l.ocache = k - 1
        return l.objs[k-1].r, int(i - l.objs[k-1].i)
 }