runtime: use TOPFRAME to identify top-of-frame functions

No change to actual runtime, but helps reduce the laundry list
of functions.

mcall, morestack, and asmcgocall are not actually top-of-frame,
so those need more attention in follow-up CLs.

mstart moved to assembly so that it can be marked TOPFRAME.

Since TOPFRAME also tells DWARF consumers not to unwind
this way, this change should also improve debuggers a
marginal amount.

This CL is part of a stack adding windows/arm64
support (#36439), intended to land in the Go 1.17 cycle.
This CL is, however, not windows/arm64-specific.
It is cleanup meant to make the port (and future ports) easier.

Change-Id: If1e0d46ca973de5e46b62948d076f675f285b5d9
Reviewed-on: https://go-review.googlesource.com/c/go/+/288802
Trust: Russ Cox <rsc@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

15 files changed, 62 insertions, 22 deletions

diff --git a/src/runtime/asm_386.s b/src/runtime/asm_386.s
index fcf74a03cf..5b0852f780 100644
--- a/src/runtime/asm_386.s
+++ b/src/runtime/asm_386.s
@@ -89,7 +89,7 @@ GLOBL _rt0_386_lib_argc<>(SB),NOPTR, $4
 DATA _rt0_386_lib_argv<>(SB)/4, $0
 GLOBL _rt0_386_lib_argv<>(SB),NOPTR, $4
 
-TEXT runtime·rt0_go(SB),NOSPLIT|NOFRAME,$0
+TEXT runtime·rt0_go(SB),NOSPLIT|NOFRAME|TOPFRAME,$0
 	// Copy arguments forward on an even stack.
 	// Users of this function jump to it, they don't call it.
 	MOVL	0(SP), AX
@@ -269,6 +269,10 @@ TEXT runtime·asminit(SB),NOSPLIT,$0-0
 	FLDCW	runtime·controlWord64(SB)
 	RET
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	CALL	runtime·mstart0(SB)
+	RET // not reached
+
 /*
  *  go-routine
  */
@@ -1307,7 +1311,7 @@ TEXT _cgo_topofstack(SB),NOSPLIT,$0
 
 // The top-most function running on a goroutine
 // returns to goexit+PCQuantum.
-TEXT runtime·goexit(SB),NOSPLIT,$0-0
+TEXT runtime·goexit(SB),NOSPLIT|TOPFRAME,$0-0
 	BYTE	$0x90	// NOP
 	CALL	runtime·goexit1(SB)	// does not return
 	// traceback from goexit1 must hit code range of goexit
diff --git a/src/runtime/asm_amd64.s b/src/runtime/asm_amd64.s
index 8ee2ac2123..a68dc72ae5 100644
--- a/src/runtime/asm_amd64.s
+++ b/src/runtime/asm_amd64.s
@@ -84,7 +84,7 @@ 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
+TEXT runtime·rt0_go(SB),NOSPLIT|TOPFRAME,$0
 	// copy arguments forward on an even stack
 	MOVQ	DI, AX		// argc
 	MOVQ	SI, BX		// argv
@@ -250,6 +250,10 @@ TEXT runtime·asminit(SB),NOSPLIT,$0-0
 	// No per-thread init.
 	RET
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	CALL	runtime·mstart0(SB)
+	RET // not reached
+
 /*
  *  go-routine
  */
@@ -1440,7 +1444,7 @@ TEXT _cgo_topofstack(SB),NOSPLIT,$0
 // 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
+TEXT runtime·goexit<ABIInternal>(SB),NOSPLIT|TOPFRAME,$0-0
 	BYTE	$0x90	// NOP
 	CALL	runtime·goexit1(SB)	// does not return
 	// traceback from goexit1 must hit code range of goexit
diff --git a/src/runtime/asm_arm.s b/src/runtime/asm_arm.s
index 4620f19074..f9535bb1bc 100644
--- a/src/runtime/asm_arm.s
+++ b/src/runtime/asm_arm.s
@@ -112,7 +112,7 @@ GLOBL _rt0_arm_lib_argv<>(SB),NOPTR,$4
 
 // using NOFRAME means do not save LR on stack.
 // argc is in R0, argv is in R1.
-TEXT runtime·rt0_go(SB),NOSPLIT|NOFRAME,$0
+TEXT runtime·rt0_go(SB),NOSPLIT|NOFRAME|TOPFRAME,$0
 	MOVW	$0xcafebabe, R12
 
 	// copy arguments forward on an even stack
@@ -202,6 +202,10 @@ TEXT runtime·asminit(SB),NOSPLIT,$0-0
 	WORD	$0xeee1ba10	// vmsr fpscr, r11
 	RET
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	BL	runtime·mstart0(SB)
+	RET // not reached
+
 /*
  *  go-routine
  */
diff --git a/src/runtime/asm_arm64.s b/src/runtime/asm_arm64.s
index 4f0a680fa4..d81759537e 100644
--- a/src/runtime/asm_arm64.s
+++ b/src/runtime/asm_arm64.s
@@ -8,7 +8,7 @@
 #include "funcdata.h"
 #include "textflag.h"
 
-TEXT runtime·rt0_go(SB),NOSPLIT,$0
+TEXT runtime·rt0_go(SB),NOSPLIT|TOPFRAME,$0
 	// SP = stack; R0 = argc; R1 = argv
 
 	SUB	$32, RSP
@@ -109,6 +109,10 @@ TEXT runtime·breakpoint(SB),NOSPLIT|NOFRAME,$0-0
 TEXT runtime·asminit(SB),NOSPLIT|NOFRAME,$0-0
 	RET
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	BL	runtime·mstart0(SB)
+	RET // not reached
+
 /*
  *  go-routine
  */
diff --git a/src/runtime/asm_mips64x.s b/src/runtime/asm_mips64x.s
index f6d8931a15..af27b9b555 100644
--- a/src/runtime/asm_mips64x.s
+++ b/src/runtime/asm_mips64x.s
@@ -11,7 +11,7 @@
 
 #define	REGCTXT	R22
 
-TEXT runtime·rt0_go(SB),NOSPLIT,$0
+TEXT runtime·rt0_go(SB),NOSPLIT|TOPFRAME,$0
 	// R29 = stack; R4 = argc; R5 = argv
 
 	ADDV	$-24, R29
@@ -85,6 +85,10 @@ TEXT runtime·breakpoint(SB),NOSPLIT|NOFRAME,$0-0
 TEXT runtime·asminit(SB),NOSPLIT|NOFRAME,$0-0
 	RET
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	JAL	runtime·mstart0(SB)
+	RET // not reached
+
 /*
  *  go-routine
  */
diff --git a/src/runtime/asm_mipsx.s b/src/runtime/asm_mipsx.s
index cf4b1b42cc..0c7d28dcf7 100644
--- a/src/runtime/asm_mipsx.s
+++ b/src/runtime/asm_mipsx.s
@@ -11,7 +11,7 @@
 
 #define	REGCTXT	R22
 
-TEXT runtime·rt0_go(SB),NOSPLIT,$0
+TEXT runtime·rt0_go(SB),NOSPLIT|TOPFRAME,$0
 	// R29 = stack; R4 = argc; R5 = argv
 
 	ADDU	$-12, R29
@@ -86,6 +86,10 @@ TEXT runtime·breakpoint(SB),NOSPLIT,$0-0
 TEXT runtime·asminit(SB),NOSPLIT,$0-0
 	RET
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	JAL	runtime·mstart0(SB)
+	RET // not reached
+
 /*
  *  go-routine
  */
diff --git a/src/runtime/asm_ppc64x.s b/src/runtime/asm_ppc64x.s
index 90f14d8e54..56e73742ea 100644
--- a/src/runtime/asm_ppc64x.s
+++ b/src/runtime/asm_ppc64x.s
@@ -16,7 +16,7 @@
 #define cgoCalleeStackSize 32
 #endif
 
-TEXT runtime·rt0_go(SB),NOSPLIT,$0
+TEXT runtime·rt0_go(SB),NOSPLIT|TOPFRAME,$0
 	// R1 = stack; R3 = argc; R4 = argv; R13 = C TLS base pointer
 
 	// initialize essential registers
@@ -124,6 +124,10 @@ TEXT runtime·reginit(SB),NOSPLIT|NOFRAME,$0-0
 	XOR R0, R0
 	RET
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	BL	runtime·mstart0(SB)
+	RET // not reached
+
 /*
  *  go-routine
  */
diff --git a/src/runtime/asm_riscv64.s b/src/runtime/asm_riscv64.s
index d06c77b948..30f2bd2e4a 100644
--- a/src/runtime/asm_riscv64.s
+++ b/src/runtime/asm_riscv64.s
@@ -7,7 +7,7 @@
 #include "textflag.h"
 
 // func rt0_go()
-TEXT runtime·rt0_go(SB),NOSPLIT,$0
+TEXT runtime·rt0_go(SB),NOSPLIT|TOPFRAME,$0
 	// X2 = stack; A0 = argc; A1 = argv
 	ADD	$-24, X2
 	MOV	A0, 8(X2)	// argc
@@ -70,6 +70,10 @@ nocgo:
 	WORD $0 // crash if reached
 	RET
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	CALL	runtime·mstart0(SB)
+	RET // not reached
+
 // void setg_gcc(G*); set g called from gcc with g in A0
 TEXT setg_gcc<>(SB),NOSPLIT,$0-0
 	MOV	A0, g
diff --git a/src/runtime/asm_s390x.s b/src/runtime/asm_s390x.s
index 203754f32c..f9fb1a4c55 100644
--- a/src/runtime/asm_s390x.s
+++ b/src/runtime/asm_s390x.s
@@ -84,7 +84,7 @@ GLOBL _rt0_s390x_lib_argc<>(SB), NOPTR, $8
 DATA _rt0_s90x_lib_argv<>(SB)/8, $0
 GLOBL _rt0_s390x_lib_argv<>(SB), NOPTR, $8
 
-TEXT runtime·rt0_go(SB),NOSPLIT,$0
+TEXT runtime·rt0_go(SB),NOSPLIT|TOPFRAME,$0
 	// R2 = argc; R3 = argv; R11 = temp; R13 = g; R15 = stack pointer
 	// C TLS base pointer in AR0:AR1
 
@@ -170,6 +170,10 @@ TEXT runtime·breakpoint(SB),NOSPLIT|NOFRAME,$0-0
 TEXT runtime·asminit(SB),NOSPLIT|NOFRAME,$0-0
 	RET
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	CALL	runtime·mstart0(SB)
+	RET // not reached
+
 /*
  *  go-routine
  */
diff --git a/src/runtime/asm_wasm.s b/src/runtime/asm_wasm.s
index 3765c756b3..33c335ba5a 100644
--- a/src/runtime/asm_wasm.s
+++ b/src/runtime/asm_wasm.s
@@ -7,7 +7,7 @@
 #include "funcdata.h"
 #include "textflag.h"
 
-TEXT runtime·rt0_go(SB), NOSPLIT|NOFRAME, $0
+TEXT runtime·rt0_go(SB), NOSPLIT|NOFRAME|TOPFRAME, $0
 	// save m->g0 = g0
 	MOVD $runtime·g0(SB), runtime·m0+m_g0(SB)
 	// save m0 to g0->m
@@ -24,6 +24,10 @@ TEXT runtime·rt0_go(SB), NOSPLIT|NOFRAME, $0
 	CALL runtime·mstart(SB) // WebAssembly stack will unwind when switching to another goroutine
 	UNDEF
 
+TEXT runtime·mstart(SB),NOSPLIT|TOPFRAME,$0
+	CALL	runtime·mstart0(SB)
+	RET // not reached
+
 DATA  runtime·mainPC+0(SB)/8,$runtime·main(SB)
 GLOBL runtime·mainPC(SB),RODATA,$8
 
@@ -424,7 +428,7 @@ CALLFN(·call268435456, 268435456)
 CALLFN(·call536870912, 536870912)
 CALLFN(·call1073741824, 1073741824)
 
-TEXT runtime·goexit(SB), NOSPLIT, $0-0
+TEXT runtime·goexit(SB), NOSPLIT|TOPFRAME, $0-0
 	NOP // first PC of goexit is skipped
 	CALL runtime·goexit1(SB) // does not return
 	UNDEF
diff --git a/src/runtime/proc.go b/src/runtime/proc.go
index e6670135db..ccfe085691 100644
--- a/src/runtime/proc.go
+++ b/src/runtime/proc.go
@@ -1234,7 +1234,10 @@ func mStackIsSystemAllocated() bool {
 }
 
 // mstart is the entry-point for new Ms.
-//
+// It is written in assembly, marked TOPFRAME, and calls mstart0.
+func mstart()
+
+// mstart0 is the Go entry-point for new Ms.
 // This must not split the stack because we may not even have stack
 // bounds set up yet.
 //
@@ -1243,7 +1246,7 @@ func mStackIsSystemAllocated() bool {
 //
 //go:nosplit
 //go:nowritebarrierrec
-func mstart() {
+func mstart0() {
 	_g_ := getg()
 
 	osStack := _g_.stack.lo == 0
diff --git a/src/runtime/sys_windows_386.s b/src/runtime/sys_windows_386.s
index 576dd28771..4f00c58c16 100644
--- a/src/runtime/sys_windows_386.s
+++ b/src/runtime/sys_windows_386.s
@@ -174,7 +174,7 @@ TEXT runtime·profileloop(SB),NOSPLIT,$0
 	ADDL	$12, SP
 	JMP	CX
 
-TEXT runtime·externalthreadhandler(SB),NOSPLIT,$0
+TEXT runtime·externalthreadhandler(SB),NOSPLIT|TOPFRAME,$0
 	PUSHL	BP
 	MOVL	SP, BP
 	PUSHL	BX
diff --git a/src/runtime/sys_windows_amd64.s b/src/runtime/sys_windows_amd64.s
index 6f2abb5444..aba2811e59 100644
--- a/src/runtime/sys_windows_amd64.s
+++ b/src/runtime/sys_windows_amd64.s
@@ -215,7 +215,7 @@ TEXT runtime·profileloop(SB),NOSPLIT|NOFRAME,$8
 	CALL	runtime·externalthreadhandler(SB)
 	RET
 
-TEXT runtime·externalthreadhandler(SB),NOSPLIT|NOFRAME,$0
+TEXT runtime·externalthreadhandler(SB),NOSPLIT|NOFRAME|TOPFRAME,$0
 	PUSHQ	BP
 	MOVQ	SP, BP
 	PUSHQ	BX
diff --git a/src/runtime/sys_windows_arm.s b/src/runtime/sys_windows_arm.s
index 7d134dd3f4..a55f474d39 100644
--- a/src/runtime/sys_windows_arm.s
+++ b/src/runtime/sys_windows_arm.s
@@ -257,7 +257,7 @@ TEXT runtime·profileloop(SB),NOSPLIT|NOFRAME,$0
 // 0 | retval         |
 //   +----------------+
 //
-TEXT runtime·externalthreadhandler(SB),NOSPLIT|NOFRAME,$0
+TEXT runtime·externalthreadhandler(SB),NOSPLIT|NOFRAME|TOPFRAME,$0
 	MOVM.DB.W [R4-R11, R14], (R13)		// push {r4-r11, lr}
 	SUB	$(m__size + g__size + 20), R13	// space for locals
 	MOVW	R0, 12(R13)
diff --git a/src/runtime/traceback.go b/src/runtime/traceback.go
index 18d8a42854..e2bd968919 100644
--- a/src/runtime/traceback.go
+++ b/src/runtime/traceback.go
@@ -1002,12 +1002,9 @@ func tracebackHexdump(stk stack, frame *stkframe, bad uintptr) {
 
 // Does f mark the top of a goroutine stack?
 func topofstack(f funcInfo, g0 bool) bool {
-	return f.funcID == funcID_goexit ||
-		f.funcID == funcID_mstart ||
+	return f.flag&funcFlag_TOPFRAME != 0 ||
 		f.funcID == funcID_mcall ||
 		f.funcID == funcID_morestack ||
-		f.funcID == funcID_rt0_go ||
-		f.funcID == funcID_externalthreadhandler ||
 		// asmcgocall is TOS on the system stack because it
 		// switches to the system stack, but in this case we
 		// can come back to the regular stack and still want