cmd/compile: use correct type for byteswaps on multi-byte stores

Use the type of the store for the byteswap, not the type of the
store's value argument.

Normally when we're storing a 16-bit value, the value being stored is
also typed as 16 bits. But sometimes it is typed as something smaller,
usually because it is the result of an upcast from a smaller value,
and that upcast needs no instructions.

If the type of the store's arg is thinner than the type being stored,
and the byteswap'd value uses that thinner type, and the byteswap'd
value needs to be spilled & restored, that spill/restore happens using
the thinner type, which causes us to lose some of the top bits of the
value.

Fixes #59367

Change-Id: If6ce1e8a76f18bf8e9d79871b6caa438bc3cce4d
Reviewed-on: https://go-review.googlesource.com/c/go/+/481395
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>

1 files changed, 80 insertions, 0 deletions

diff --git a/test/fixedbugs/issue59367.go b/test/fixedbugs/issue59367.go
new file mode 100644
index 0000000000..32a79e1208
--- /dev/null
+++ b/test/fixedbugs/issue59367.go
@@ -0,0 +1,80 @@
+// run
+
+// Copyright 2023 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package main
+
+func main() {
+	var b [8]byte
+	one := uint8(1)
+	f16(&one, b[:2])
+	if b[1] != 1 {
+		println("2-byte value lost")
+	}
+	f32(&one, b[:4])
+	if b[3] != 1 {
+		println("4-byte value lost")
+	}
+	f64(&one, b[:8])
+	if b[7] != 1 {
+		println("8-byte value lost")
+	}
+}
+
+//go:noinline
+func f16(p *uint8, b []byte) {
+	_ = b[1]            // bounds check
+	x := *p             // load a byte
+	y := uint16(x)      // zero extend to 16 bits
+	b[0] = byte(y >> 8) // compute ROLW
+	b[1] = byte(y)
+	nop()               // spill/restore ROLW
+	b[0] = byte(y >> 8) // use ROLW
+	b[1] = byte(y)
+}
+
+//go:noinline
+func f32(p *uint8, b []byte) {
+	_ = b[3]             // bounds check
+	x := *p              // load a byte
+	y := uint32(x)       // zero extend to 32 bits
+	b[0] = byte(y >> 24) // compute ROLL
+	b[1] = byte(y >> 16)
+	b[2] = byte(y >> 8)
+	b[3] = byte(y)
+	nop()                // spill/restore ROLL
+	b[0] = byte(y >> 24) // use ROLL
+	b[1] = byte(y >> 16)
+	b[2] = byte(y >> 8)
+	b[3] = byte(y)
+}
+
+//go:noinline
+func f64(p *uint8, b []byte) {
+	_ = b[7]             // bounds check
+	x := *p              // load a byte
+	y := uint64(x)       // zero extend to 64 bits
+	b[0] = byte(y >> 56) // compute ROLQ
+	b[1] = byte(y >> 48)
+	b[2] = byte(y >> 40)
+	b[3] = byte(y >> 32)
+	b[4] = byte(y >> 24)
+	b[5] = byte(y >> 16)
+	b[6] = byte(y >> 8)
+	b[7] = byte(y)
+	nop()                // spill/restore ROLQ
+	b[0] = byte(y >> 56) // use ROLQ
+	b[1] = byte(y >> 48)
+	b[2] = byte(y >> 40)
+	b[3] = byte(y >> 32)
+	b[4] = byte(y >> 24)
+	b[5] = byte(y >> 16)
+	b[6] = byte(y >> 8)
+	b[7] = byte(y)
+}
+
+//go:noinline
+func nop() {
+}