cmd/compile: de-virtualize interface calls

With this change, code like

    h := sha1.New()
    h.Write(buf)
    sum := h.Sum()

gets compiled into static calls rather than
interface calls, because the compiler is able
to prove that 'h' is really a *sha1.digest.

The InterCall re-write rule hits a few dozen times
during make.bash, and hundreds of times during all.bash.

The most common pattern identified by the compiler
is a constructor like

    func New() Interface { return &impl{...} }

where the constructor gets inlined into the caller,
and the result is used immediately. Examples include
{sha1,md5,crc32,crc64,...}.New, base64.NewEncoder,
base64.NewDecoder, errors.New, net.Pipe, and so on.

Some existing benchmarks that change on darwin/amd64:

Crc64/ISO4KB-8        2.67µs ± 1%    2.66µs ± 0%  -0.36%  (p=0.015 n=10+10)
Crc64/ISO1KB-8         694ns ± 0%     690ns ± 1%  -0.59%  (p=0.001 n=10+10)
Adler32KB-8            473ns ± 1%     471ns ± 0%  -0.39%  (p=0.010 n=10+9)

On architectures like amd64, the reduction in code size
appears to contribute more to benchmark improvements than just
removing the indirect call, since that branch gets predicted
accurately when called in a loop.

Updates #19361

Change-Id: Ia9d30afdd5f6b4d38d38b14b88f308acae8ce7ed
Reviewed-on: https://go-review.googlesource.com/37751
Run-TryBot: Philip Hofer <phofer@umich.edu>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

1 files changed, 64 insertions, 0 deletions

diff --git a/test/devirt.go b/test/devirt.go
new file mode 100644
index 0000000000..a2211f185c
--- /dev/null
+++ b/test/devirt.go
@@ -0,0 +1,64 @@
+// errorcheck -0 -d=ssa/opt/debug=3
+
+package main
+
+import (
+	"crypto/sha1"
+	"errors"
+	"fmt"
+	"sync"
+)
+
+func f0() {
+	v := errors.New("error string")
+	_ = v.Error() // ERROR "de-virtualizing call$"
+}
+
+func f1() {
+	h := sha1.New()
+	buf := make([]byte, 4)
+	h.Write(buf)   // ERROR "de-virtualizing call$"
+	_ = h.Sum(nil) // ERROR "de-virtualizing call$"
+}
+
+func f2() {
+	// trickier case: make sure we see this is *sync.rlocker
+	// instead of *sync.RWMutex,
+	// even though they are the same pointers
+	var m sync.RWMutex
+	r := m.RLocker()
+
+	// deadlock if the type of 'r' is improperly interpreted
+	// as *sync.RWMutex
+	r.Lock() // ERROR "de-virtualizing call$"
+	m.RLock()
+	r.Unlock() // ERROR "de-virtualizing call$"
+	m.RUnlock()
+}
+
+type multiword struct{ a, b, c int }
+
+func (m multiword) Error() string { return fmt.Sprintf("%d, %d, %d", m.a, m.b, m.c) }
+
+func f3() {
+	// can't de-virtualize this one yet;
+	// it passes through a call to iconvT2I
+	var err error
+	err = multiword{1, 2, 3}
+	if err.Error() != "1, 2, 3" {
+		panic("bad call")
+	}
+
+	// ... but we can do this one
+	err = &multiword{1, 2, 3}
+	if err.Error() != "1, 2, 3" { // ERROR "de-virtualizing call$"
+		panic("bad call")
+	}
+}
+
+func main() {
+	f0()
+	f1()
+	f2()
+	f3()
+}