runtime: do not trace past jmpdefer during pprof traceback on arm

jmpdefer modifies PC, SP, and LR, and not atomically,
so walking past jmpdefer will often end up in a state
where the three are not a consistent execution snapshot.
This was causing warning messages a few frames later
when the traceback realized it was confused, but given
the right memory it could easily crash instead.

Update #8153

LGTM=minux, iant
R=golang-codereviews, minux, iant
CC=golang-codereviews, r
https://golang.org/cl/107970043

src/pkg/runtime/asm_arm.s

@@ -394,6 +394,10 @@ TEXT runtime·lessstack(SB), NOSPLIT, $-4-0
 // 1. grab stored LR for caller
 // 2. sub 4 bytes to get back to BL deferreturn
 // 3. B to fn
+// TODO(rsc): Push things on stack and then use pop
+// to load all registers simultaneously, so that a profiling
+// interrupt can never see mismatched SP/LR/PC.
+// (And double-check that pop is atomic in that way.)
 TEXT runtime·jmpdefer(SB), NOSPLIT, $0-8
 	MOVW	0(SP), LR
 	MOVW	$-4(LR), LR	// BL deferreturn

src/pkg/runtime/traceback_arm.c

@@ -110,6 +110,19 @@ runtime·gentraceback(uintptr pc0, uintptr sp0, uintptr lr0, G *gp, int32 skip,
 		if(runtime·topofstack(f)) {
 			frame.lr = 0;
 			flr = nil;
+		} else if(f->entry == (uintptr)runtime·jmpdefer) {
+			// jmpdefer modifies SP/LR/PC non-atomically.
+			// If a profiling interrupt arrives during jmpdefer,
+			// the stack unwind may see a mismatched register set
+			// and get confused. Stop if we see PC within jmpdefer
+			// to avoid that confusion.
+			// See golang.org/issue/8153.
+			// This check can be deleted if jmpdefer is changed
+			// to restore all three atomically using pop.
+			if(callback != nil)
+				runtime·throw("traceback_arm: found jmpdefer when tracing with callback");
+			frame.lr = 0;
+			flr = nil;
 		} else {
 			if((n == 0 && frame.sp < frame.fp) || frame.lr == 0)
 				frame.lr = *(uintptr*)frame.sp;