internal/sys: add LR and fixed frame size to sys.Arch

Storing this information in the Arch eliminates some code duplication
between the compiler and linker. This information is entirely
determined by the Arch, so the current approach of attaching it to an
entire Ctxt is a little silly. This will also make it easier to use
this information from tests.

The next CL will be a rote refactoring to eliminate the
Ctxt.FixedFrameSize methods.

Change-Id: I315c524fa66a0ea99f63ae5a2a6fdc367d843bad
Reviewed-on: https://go-review.googlesource.com/c/go/+/400818
Run-TryBot: Austin Clements <austin@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
This commit is contained in:
Austin Clements 2022-04-18 13:39:52 -04:00
parent caa46312ee
commit 8af7fac5a1
5 changed files with 152 additions and 133 deletions

View File

@ -989,16 +989,7 @@ func (fi *FuncInfo) UnspillRegisterArgs(last *Prog, pa ProgAlloc) *Prog {
// on the stack in the function prologue and so always have a pointer between
// the hardware stack pointer and the local variable area.
func (ctxt *Link) FixedFrameSize() int64 {
switch ctxt.Arch.Family {
case sys.AMD64, sys.I386, sys.Wasm:
return 0
case sys.PPC64:
// PIC code on ppc64le requires 32 bytes of stack, and it's easier to
// just use that much stack always on ppc64x.
return int64(4 * ctxt.Arch.PtrSize)
default:
return int64(ctxt.Arch.PtrSize)
}
return ctxt.Arch.FixedFrameSize
}
// LinkArch is the definition of a single architecture.

View File

@ -56,6 +56,18 @@ type Arch struct {
// CanJumpTable reports whether the backend can handle
// compiling a jump table.
CanJumpTable bool
// HasLR indicates that this architecture uses a link register
// for calls.
HasLR bool
// FixedFrameSize is the smallest possible offset from the
// hardware stack pointer to a local variable on the stack.
// Architectures that use a link register save its value on
// the stack in the function prologue and so always have a
// pointer between the hardware stack pointer and the local
// variable area.
FixedFrameSize int64
}
// InFamily reports whether a is a member of any of the specified
@ -78,6 +90,8 @@ var Arch386 = &Arch{
MinLC: 1,
Alignment: 1,
CanMergeLoads: true,
HasLR: false,
FixedFrameSize: 0,
}
var ArchAMD64 = &Arch{
@ -90,6 +104,8 @@ var ArchAMD64 = &Arch{
Alignment: 1,
CanMergeLoads: true,
CanJumpTable: true,
HasLR: false,
FixedFrameSize: 0,
}
var ArchARM = &Arch{
@ -101,6 +117,8 @@ var ArchARM = &Arch{
MinLC: 4,
Alignment: 4, // TODO: just for arm5?
CanMergeLoads: false,
HasLR: true,
FixedFrameSize: 4, // LR
}
var ArchARM64 = &Arch{
@ -112,6 +130,8 @@ var ArchARM64 = &Arch{
MinLC: 4,
Alignment: 1,
CanMergeLoads: true,
HasLR: true,
FixedFrameSize: 8, // LR
}
var ArchLoong64 = &Arch{
@ -123,6 +143,8 @@ var ArchLoong64 = &Arch{
MinLC: 4,
Alignment: 8, // Unaligned accesses are not guaranteed to be fast
CanMergeLoads: false,
HasLR: true,
FixedFrameSize: 8, // LR
}
var ArchMIPS = &Arch{
@ -134,6 +156,8 @@ var ArchMIPS = &Arch{
MinLC: 4,
Alignment: 4,
CanMergeLoads: false,
HasLR: true,
FixedFrameSize: 4, // LR
}
var ArchMIPSLE = &Arch{
@ -145,6 +169,8 @@ var ArchMIPSLE = &Arch{
MinLC: 4,
Alignment: 4,
CanMergeLoads: false,
HasLR: true,
FixedFrameSize: 4, // LR
}
var ArchMIPS64 = &Arch{
@ -156,6 +182,8 @@ var ArchMIPS64 = &Arch{
MinLC: 4,
Alignment: 8,
CanMergeLoads: false,
HasLR: true,
FixedFrameSize: 8, // LR
}
var ArchMIPS64LE = &Arch{
@ -167,6 +195,8 @@ var ArchMIPS64LE = &Arch{
MinLC: 4,
Alignment: 8,
CanMergeLoads: false,
HasLR: true,
FixedFrameSize: 8, // LR
}
var ArchPPC64 = &Arch{
@ -178,6 +208,10 @@ var ArchPPC64 = &Arch{
MinLC: 4,
Alignment: 1,
CanMergeLoads: false,
HasLR: true,
// PIC code on ppc64le requires 32 bytes of stack, and it's
// easier to just use that much stack always.
FixedFrameSize: 4 * 8,
}
var ArchPPC64LE = &Arch{
@ -189,6 +223,8 @@ var ArchPPC64LE = &Arch{
MinLC: 4,
Alignment: 1,
CanMergeLoads: true,
HasLR: true,
FixedFrameSize: 4 * 8,
}
var ArchRISCV64 = &Arch{
@ -200,6 +236,8 @@ var ArchRISCV64 = &Arch{
MinLC: 4,
Alignment: 8, // riscv unaligned loads work, but are really slow (trap + simulated by OS)
CanMergeLoads: false,
HasLR: true,
FixedFrameSize: 8, // LR
}
var ArchS390X = &Arch{
@ -211,6 +249,8 @@ var ArchS390X = &Arch{
MinLC: 2,
Alignment: 1,
CanMergeLoads: true,
HasLR: true,
FixedFrameSize: 8, // LR
}
var ArchWasm = &Arch{
@ -222,6 +262,8 @@ var ArchWasm = &Arch{
MinLC: 1,
Alignment: 1,
CanMergeLoads: false,
HasLR: false,
FixedFrameSize: 0,
}
var Archs = [...]*Arch{

View File

@ -1360,7 +1360,7 @@ func (d *dwctxt) writeframes(fs loader.Sym) dwarfSecInfo {
fsu := d.ldr.MakeSymbolUpdater(fs)
fsu.SetType(sym.SDWARFSECT)
isdw64 := isDwarf64(d.linkctxt)
haslr := haslinkregister(d.linkctxt)
haslr := d.linkctxt.Arch.HasLR
// Length field is 4 bytes on Dwarf32 and 12 bytes on Dwarf64
lengthFieldSize := int64(4)

View File

@ -2349,12 +2349,8 @@ type chain struct {
limit int // limit on entry to sym
}
func haslinkregister(ctxt *Link) bool {
return ctxt.FixedFrameSize() != 0
}
func callsize(ctxt *Link) int {
if haslinkregister(ctxt) {
if ctxt.Arch.HasLR {
return 0
}
return ctxt.Arch.RegSize
@ -2554,7 +2550,7 @@ func (sc *stkChk) print(ch *chain, limit int) {
}
} else {
sc.print(ch.up, ch.limit+callsize(ctxt))
if !haslinkregister(ctxt) {
if !ctxt.Arch.HasLR {
fmt.Printf("\t%d\ton entry to %s\n", ch.limit, name)
}
}

View File

@ -33,7 +33,6 @@ package ld
import (
"bufio"
"cmd/internal/objabi"
"cmd/internal/sys"
"cmd/link/internal/loader"
"cmd/link/internal/sym"
"debug/elf"
@ -108,16 +107,7 @@ type cgodata struct {
// on the stack in the function prologue and so always have a pointer between
// the hardware stack pointer and the local variable area.
func (ctxt *Link) FixedFrameSize() int64 {
switch ctxt.Arch.Family {
case sys.AMD64, sys.I386:
return 0
case sys.PPC64:
// PIC code on ppc64le requires 32 bytes of stack, and it's easier to
// just use that much stack always on ppc64x.
return int64(4 * ctxt.Arch.PtrSize)
default:
return int64(ctxt.Arch.PtrSize)
}
return ctxt.Arch.FixedFrameSize
}
func (ctxt *Link) Logf(format string, args ...interface{}) {