reflect,runtime: use internal ABI for selected ASM routines

Change the definitions of selected runtime assembly routines
from ABI0 (the default) to ABIInternal. The ABIInternal def is
intended to indicate that these functions don't follow the existing Go
runtime ABI. In addition, convert the assembly reference to
runtime.main (from runtime.mainPC) to ABIInternal. Finally, for
functions such as "runtime.duffzero" that are called directly from
generated code, make sure that the compiler looks up the correct
ABI version.

This is intended to support the register abi work, however these
changes should not have any issues even when GOEXPERIMENT=regabi is
not in effect.

Updates #27539, #40724.

Change-Id: I9846f8dcaccc95718cf2e61a18b7e924a0677e4c
Reviewed-on: https://go-review.googlesource.com/c/go/+/262319
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Than McIntosh <thanm@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

1 files changed, 58 insertions, 38 deletions

diff --git a/src/runtime/asm_amd64.s b/src/runtime/asm_amd64.s
index 19a3bb2d7d..196252e1dd 100644
--- a/src/runtime/asm_amd64.s
+++ b/src/runtime/asm_amd64.s
@@ -84,7 +84,9 @@ GLOBL _rt0_amd64_lib_argc<>(SB),NOPTR, $8
 DATA _rt0_amd64_lib_argv<>(SB)/8, $0
 GLOBL _rt0_amd64_lib_argv<>(SB),NOPTR, $8
 
-TEXT runtime·rt0_go(SB),NOSPLIT,$0
+// Defined as ABIInternal since it does not use the stack-based Go ABI (and
+// in addition there are no calls to this entry point from Go code).
+TEXT runtime·rt0_go<ABIInternal>(SB),NOSPLIT,$0
 	// copy arguments forward on an even stack
 	MOVQ	DI, AX		// argc
 	MOVQ	SI, BX		// argv
@@ -229,10 +231,13 @@ ok:
 
 	// Prevent dead-code elimination of debugCallV1, which is
 	// intended to be called by debuggers.
-	MOVQ	$runtime·debugCallV1(SB), AX
+	MOVQ	$runtime·debugCallV1<ABIInternal>(SB), AX
 	RET
 
-DATA	runtime·mainPC+0(SB)/8,$runtime·main(SB)
+// mainPC is a function value for runtime.main, to be passed to newproc.
+// The reference to runtime.main is made via ABIInternal, since the
+// actual function (not the ABI0 wrapper) is needed by newproc.
+DATA	runtime·mainPC+0(SB)/8,$runtime·main<ABIInternal>(SB)
 GLOBL	runtime·mainPC(SB),RODATA,$8
 
 TEXT runtime·breakpoint(SB),NOSPLIT,$0-0
@@ -468,7 +473,7 @@ TEXT runtime·morestack_noctxt(SB),NOSPLIT,$0
 	JMP	AX
 // Note: can't just "JMP NAME(SB)" - bad inlining results.
 
-TEXT ·reflectcall(SB), NOSPLIT, $0-32
+TEXT ·reflectcall<ABIInternal>(SB), NOSPLIT, $0-32
 	MOVLQZX argsize+24(FP), CX
 	DISPATCH(runtime·call16, 16)
 	DISPATCH(runtime·call32, 32)
@@ -1354,8 +1359,11 @@ TEXT _cgo_topofstack(SB),NOSPLIT,$0
 	RET
 
 // The top-most function running on a goroutine
-// returns to goexit+PCQuantum.
-TEXT runtime·goexit(SB),NOSPLIT,$0-0
+// returns to goexit+PCQuantum. Defined as ABIInternal
+// so as to make it identifiable to traceback (this
+// function it used as a sentinel; traceback wants to
+// see the func PC, not a wrapper PC).
+TEXT runtime·goexit<ABIInternal>(SB),NOSPLIT,$0-0
 	BYTE	$0x90	// NOP
 	CALL	runtime·goexit1(SB)	// does not return
 	// traceback from goexit1 must hit code range of goexit
@@ -1377,7 +1385,8 @@ TEXT runtime·addmoduledata(SB),NOSPLIT,$0-0
 // - AX is the value being written at DI
 // It clobbers FLAGS. It does not clobber any general-purpose registers,
 // but may clobber others (e.g., SSE registers).
-TEXT runtime·gcWriteBarrier(SB),NOSPLIT,$120
+// Defined as ABIInternal since it does not use the stack-based Go ABI.
+TEXT runtime·gcWriteBarrier<ABIInternal>(SB),NOSPLIT,$120
 	// Save the registers clobbered by the fast path. This is slightly
 	// faster than having the caller spill these.
 	MOVQ	R14, 104(SP)
@@ -1461,51 +1470,58 @@ flush:
 	JMP	ret
 
 // gcWriteBarrierCX is gcWriteBarrier, but with args in DI and CX.
-TEXT runtime·gcWriteBarrierCX(SB),NOSPLIT,$0
+// Defined as ABIInternal since it does not use the stable Go ABI.
+TEXT runtime·gcWriteBarrierCX<ABIInternal>(SB),NOSPLIT,$0
 	XCHGQ CX, AX
-	CALL runtime·gcWriteBarrier(SB)
+	CALL runtime·gcWriteBarrier<ABIInternal>(SB)
 	XCHGQ CX, AX
 	RET
 
 // gcWriteBarrierDX is gcWriteBarrier, but with args in DI and DX.
-TEXT runtime·gcWriteBarrierDX(SB),NOSPLIT,$0
+// Defined as ABIInternal since it does not use the stable Go ABI.
+TEXT runtime·gcWriteBarrierDX<ABIInternal>(SB),NOSPLIT,$0
 	XCHGQ DX, AX
-	CALL runtime·gcWriteBarrier(SB)
+	CALL runtime·gcWriteBarrier<ABIInternal>(SB)
 	XCHGQ DX, AX
 	RET
 
 // gcWriteBarrierBX is gcWriteBarrier, but with args in DI and BX.
-TEXT runtime·gcWriteBarrierBX(SB),NOSPLIT,$0
+// Defined as ABIInternal since it does not use the stable Go ABI.
+TEXT runtime·gcWriteBarrierBX<ABIInternal>(SB),NOSPLIT,$0
 	XCHGQ BX, AX
-	CALL runtime·gcWriteBarrier(SB)
+	CALL runtime·gcWriteBarrier<ABIInternal>(SB)
 	XCHGQ BX, AX
 	RET
 
 // gcWriteBarrierBP is gcWriteBarrier, but with args in DI and BP.
-TEXT runtime·gcWriteBarrierBP(SB),NOSPLIT,$0
+// Defined as ABIInternal since it does not use the stable Go ABI.
+TEXT runtime·gcWriteBarrierBP<ABIInternal>(SB),NOSPLIT,$0
 	XCHGQ BP, AX
-	CALL runtime·gcWriteBarrier(SB)
+	CALL runtime·gcWriteBarrier<ABIInternal>(SB)
 	XCHGQ BP, AX
 	RET
 
 // gcWriteBarrierSI is gcWriteBarrier, but with args in DI and SI.
-TEXT runtime·gcWriteBarrierSI(SB),NOSPLIT,$0
+// Defined as ABIInternal since it does not use the stable Go ABI.
+TEXT runtime·gcWriteBarrierSI<ABIInternal>(SB),NOSPLIT,$0
 	XCHGQ SI, AX
-	CALL runtime·gcWriteBarrier(SB)
+	CALL runtime·gcWriteBarrier<ABIInternal>(SB)
 	XCHGQ SI, AX
 	RET
 
 // gcWriteBarrierR8 is gcWriteBarrier, but with args in DI and R8.
-TEXT runtime·gcWriteBarrierR8(SB),NOSPLIT,$0
+// Defined as ABIInternal since it does not use the stable Go ABI.
+TEXT runtime·gcWriteBarrierR8<ABIInternal>(SB),NOSPLIT,$0
 	XCHGQ R8, AX
-	CALL runtime·gcWriteBarrier(SB)
+	CALL runtime·gcWriteBarrier<ABIInternal>(SB)
 	XCHGQ R8, AX
 	RET
 
 // gcWriteBarrierR9 is gcWriteBarrier, but with args in DI and R9.
-TEXT runtime·gcWriteBarrierR9(SB),NOSPLIT,$0
+// Defined as ABIInternal since it does not use the stable Go ABI.
+TEXT runtime·gcWriteBarrierR9<ABIInternal>(SB),NOSPLIT,$0
 	XCHGQ R9, AX
-	CALL runtime·gcWriteBarrier(SB)
+	CALL runtime·gcWriteBarrier<ABIInternal>(SB)
 	XCHGQ R9, AX
 	RET
 
@@ -1544,7 +1560,10 @@ GLOBL	debugCallFrameTooLarge<>(SB), RODATA, $20	// Size duplicated below
 // obey escape analysis requirements. Specifically, it must not pass
 // a stack pointer to an escaping argument. debugCallV1 cannot check
 // this invariant.
-TEXT runtime·debugCallV1(SB),NOSPLIT,$152-0
+//
+// This is ABIInternal because Go code injects its PC directly into new
+// goroutine stacks.
+TEXT runtime·debugCallV1<ABIInternal>(SB),NOSPLIT,$152-0
 	// Save all registers that may contain pointers so they can be
 	// conservatively scanned.
 	//
@@ -1705,67 +1724,68 @@ TEXT runtime·debugCallPanicked(SB),NOSPLIT,$16-16
 // in the caller's stack frame. These stubs write the args into that stack space and
 // then tail call to the corresponding runtime handler.
 // The tail call makes these stubs disappear in backtraces.
-TEXT runtime·panicIndex(SB),NOSPLIT,$0-16
+// Defined as ABIInternal since they do not use the stack-based Go ABI.
+TEXT runtime·panicIndex<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	AX, x+0(FP)
 	MOVQ	CX, y+8(FP)
 	JMP	runtime·goPanicIndex(SB)
-TEXT runtime·panicIndexU(SB),NOSPLIT,$0-16
+TEXT runtime·panicIndexU<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	AX, x+0(FP)
 	MOVQ	CX, y+8(FP)
 	JMP	runtime·goPanicIndexU(SB)
-TEXT runtime·panicSliceAlen(SB),NOSPLIT,$0-16
+TEXT runtime·panicSliceAlen<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	CX, x+0(FP)
 	MOVQ	DX, y+8(FP)
 	JMP	runtime·goPanicSliceAlen(SB)
-TEXT runtime·panicSliceAlenU(SB),NOSPLIT,$0-16
+TEXT runtime·panicSliceAlenU<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	CX, x+0(FP)
 	MOVQ	DX, y+8(FP)
 	JMP	runtime·goPanicSliceAlenU(SB)
-TEXT runtime·panicSliceAcap(SB),NOSPLIT,$0-16
+TEXT runtime·panicSliceAcap<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	CX, x+0(FP)
 	MOVQ	DX, y+8(FP)
 	JMP	runtime·goPanicSliceAcap(SB)
-TEXT runtime·panicSliceAcapU(SB),NOSPLIT,$0-16
+TEXT runtime·panicSliceAcapU<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	CX, x+0(FP)
 	MOVQ	DX, y+8(FP)
 	JMP	runtime·goPanicSliceAcapU(SB)
-TEXT runtime·panicSliceB(SB),NOSPLIT,$0-16
+TEXT runtime·panicSliceB<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	AX, x+0(FP)
 	MOVQ	CX, y+8(FP)
 	JMP	runtime·goPanicSliceB(SB)
-TEXT runtime·panicSliceBU(SB),NOSPLIT,$0-16
+TEXT runtime·panicSliceBU<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	AX, x+0(FP)
 	MOVQ	CX, y+8(FP)
 	JMP	runtime·goPanicSliceBU(SB)
-TEXT runtime·panicSlice3Alen(SB),NOSPLIT,$0-16
+TEXT runtime·panicSlice3Alen<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	DX, x+0(FP)
 	MOVQ	BX, y+8(FP)
 	JMP	runtime·goPanicSlice3Alen(SB)
-TEXT runtime·panicSlice3AlenU(SB),NOSPLIT,$0-16
+TEXT runtime·panicSlice3AlenU<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	DX, x+0(FP)
 	MOVQ	BX, y+8(FP)
 	JMP	runtime·goPanicSlice3AlenU(SB)
-TEXT runtime·panicSlice3Acap(SB),NOSPLIT,$0-16
+TEXT runtime·panicSlice3Acap<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	DX, x+0(FP)
 	MOVQ	BX, y+8(FP)
 	JMP	runtime·goPanicSlice3Acap(SB)
-TEXT runtime·panicSlice3AcapU(SB),NOSPLIT,$0-16
+TEXT runtime·panicSlice3AcapU<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	DX, x+0(FP)
 	MOVQ	BX, y+8(FP)
 	JMP	runtime·goPanicSlice3AcapU(SB)
-TEXT runtime·panicSlice3B(SB),NOSPLIT,$0-16
+TEXT runtime·panicSlice3B<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	CX, x+0(FP)
 	MOVQ	DX, y+8(FP)
 	JMP	runtime·goPanicSlice3B(SB)
-TEXT runtime·panicSlice3BU(SB),NOSPLIT,$0-16
+TEXT runtime·panicSlice3BU<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	CX, x+0(FP)
 	MOVQ	DX, y+8(FP)
 	JMP	runtime·goPanicSlice3BU(SB)
-TEXT runtime·panicSlice3C(SB),NOSPLIT,$0-16
+TEXT runtime·panicSlice3C<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	AX, x+0(FP)
 	MOVQ	CX, y+8(FP)
 	JMP	runtime·goPanicSlice3C(SB)
-TEXT runtime·panicSlice3CU(SB),NOSPLIT,$0-16
+TEXT runtime·panicSlice3CU<ABIInternal>(SB),NOSPLIT,$0-16
 	MOVQ	AX, x+0(FP)
 	MOVQ	CX, y+8(FP)
 	JMP	runtime·goPanicSlice3CU(SB)