go/test/abi/part_live_2.go
David Chase 8462169b5a cmd/compile: pre-spill pointers in aggregate-typed register args
There's a problem in liveness, where liveness of any
part of an aggregate keeps the whole aggregate alive,
but the not-live parts don't get spilled.  The GC
can observe those live-but-not-spilled slots, which
can contain junk.

A better fix is to change liveness to work
pointer-by-pointer, but that is also a riskier,
trickier fix.

To avoid this, in the case of

(1) an aggregate input parameter
(2) containing pointers
(3) passed in registers

pre-spill the pointers.

Updates #40724.

Change-Id: I6beb8e0a353b1ae3c68c16072f56698061922c04
Reviewed-on: https://go-review.googlesource.com/c/go/+/307909
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
2021-04-07 03:42:11 +00:00

54 lines
986 B
Go

// run
// Copyright 2021 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.
// A test for partial liveness / partial spilling / compiler-induced GC failure
package main
import "runtime"
import "unsafe"
//go:registerparams
func F(s []int) {
for i, x := range s {
G(i, x)
}
GC()
H(&s[0]) // It's possible that this will make the spill redundant, but there's a bug in spill slot allocation.
G(len(s), cap(s))
GC()
}
//go:noinline
//go:registerparams
func G(int, int) {}
//go:noinline
//go:registerparams
func H(*int) {}
//go:registerparams
func GC() { runtime.GC(); runtime.GC() }
func main() {
s := make([]int, 3)
escape(s)
p := int(uintptr(unsafe.Pointer(&s[2])) + 42) // likely point to unallocated memory
poison([3]int{p, p, p})
F(s)
}
//go:noinline
//go:registerparams
func poison([3]int) {}
//go:noinline
//go:registerparams
func escape(s []int) {
g = s
}
var g []int