diff options
| author | Keith Randall <khr@golang.org> | 2025-05-17 15:05:56 -0700 |
|---|---|---|
| committer | Keith Randall <khr@golang.org> | 2025-10-06 14:11:41 -0700 |
| commit | 719dfcf8a8478d70360bf3c34c0e920be7b32994 (patch) | |
| tree | d58aaf3289de3bb18901e34b336da46b425f8075 /src/cmd/internal/obj/arm64/asm7.go | |
| parent | f3312124c2370c2f64a7f9ad29732ec30209647a (diff) | |
| download | go-719dfcf8a8478d70360bf3c34c0e920be7b32994.tar.xz | |
cmd/compile: redo arm64 LR/FP save and restore
Instead of storing LR (the return address) at 0(SP) and the FP
(parent's frame pointer) at -8(SP), store them at framesize-8(SP)
and framesize-16(SP), respectively.
We push and pop data onto the stack such that we're never accessing
anything below SP.
The prolog/epilog lengths are unchanged (3 insns for a typical prolog,
2 for a typical epilog).
We use 8 bytes more per frame.
Typical prologue:
STP.W (FP, LR), -16(SP)
MOVD SP, FP
SUB $C, SP
Typical epilogue:
ADD $C, SP
LDP.P 16(SP), (FP, LR)
RET
The previous word where we stored LR, at 0(SP), is now unused.
We could repurpose that slot for storing a local variable.
The new prolog and epilog instructions are recognized by libunwind,
so pc-sampling tools like perf should now be accurate. (TODO: except
maybe after the first RET instruction? Have to look into that.)
Update #73753 (fixes, for arm64)
Update #57302 (Quim thinks this will help on that issue)
Change-Id: I4800036a9a9a08aaaf35d9f99de79a36cf37ebb8
Reviewed-on: https://go-review.googlesource.com/c/go/+/674615
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Diffstat (limited to 'src/cmd/internal/obj/arm64/asm7.go')
| -rw-r--r-- | src/cmd/internal/obj/arm64/asm7.go | 12 |
1 files changed, 5 insertions, 7 deletions
diff --git a/src/cmd/internal/obj/arm64/asm7.go b/src/cmd/internal/obj/arm64/asm7.go index 743d09a319..281d705a3e 100644 --- a/src/cmd/internal/obj/arm64/asm7.go +++ b/src/cmd/internal/obj/arm64/asm7.go @@ -51,7 +51,6 @@ type ctxt7 struct { blitrl *obj.Prog elitrl *obj.Prog autosize int32 - extrasize int32 instoffset int64 pc int64 pool struct { @@ -1122,8 +1121,7 @@ func span7(ctxt *obj.Link, cursym *obj.LSym, newprog obj.ProgAlloc) { ctxt.Diag("arm64 ops not initialized, call arm64.buildop first") } - c := ctxt7{ctxt: ctxt, newprog: newprog, cursym: cursym, autosize: int32(p.To.Offset & 0xffffffff), extrasize: int32(p.To.Offset >> 32)} - p.To.Offset &= 0xffffffff // extrasize is no longer needed + c := ctxt7{ctxt: ctxt, newprog: newprog, cursym: cursym, autosize: int32(p.To.Offset)} // Process literal pool and allocate initial program counter for each Prog, before // generating branch veneers. @@ -2119,8 +2117,8 @@ func (c *ctxt7) aclass(a *obj.Addr) int { // a.Offset is still relative to pseudo-SP. a.Reg = obj.REG_NONE } - // The frame top 8 or 16 bytes are for FP - c.instoffset = int64(c.autosize) + a.Offset - int64(c.extrasize) + // The frame top 16 bytes are for LR/FP + c.instoffset = int64(c.autosize) + a.Offset - extrasize return autoclass(c.instoffset) case obj.NAME_PARAM: @@ -2180,8 +2178,8 @@ func (c *ctxt7) aclass(a *obj.Addr) int { // a.Offset is still relative to pseudo-SP. a.Reg = obj.REG_NONE } - // The frame top 8 or 16 bytes are for FP - c.instoffset = int64(c.autosize) + a.Offset - int64(c.extrasize) + // The frame top 16 bytes are for LR/FP + c.instoffset = int64(c.autosize) + a.Offset - extrasize case obj.NAME_PARAM: if a.Reg == REGSP { |