runtime, cmd/gc, cmd/ld: ignore method wrappers in recover

Bug #1:

Issue 5406 identified an interesting case:
        defer iface.M()
may end up calling a wrapper that copies an indirect receiver
from the iface value and then calls the real M method. That's
two calls down, not just one, and so recover() == nil always
in the real M method, even during a panic.

[For the purposes of this entire discussion, a wrapper's
implementation is a function containing an ordinary call, not
the optimized tail call form that is somtimes possible. The
tail call does not create a second frame, so it is already
handled correctly.]

Fix this bug by introducing g->panicwrap, which counts the
number of bytes on current stack segment that are due to
wrapper calls that should not count against the recover
check. All wrapper functions must now adjust g->panicwrap up
on entry and back down on exit. This adds slightly to their
expense; on the x86 it is a single instruction at entry and
exit; on the ARM it is three. However, the alternative is to
make a call to recover depend on being able to walk the stack,
which I very much want to avoid. We have enough problems
walking the stack for garbage collection and profiling.
Also, if performance is critical in a specific case, it is already
faster to use a pointer receiver and avoid this kind of wrapper
entirely.

Bug #2:

The old code, which did not consider the possibility of two
calls, already contained a check to see if the call had split
its stack and so the panic-created segment was one behind the
current segment. In the wrapper case, both of the two calls
might split their stacks, so the panic-created segment can be
two behind the current segment.

Fix this by propagating the Stktop.panic flag forward during
stack splits instead of looking backward during recover.

Fixes issue 5406.

R=golang-dev, iant
CC=golang-dev
https://codereview.appspot.com/13367052

1 files changed, 28 insertions, 4 deletions

diff --git a/src/cmd/8l/pass.c b/src/cmd/8l/pass.c
index b558ffaa9..1eaf78fe0 100644
--- a/src/cmd/8l/pass.c
+++ b/src/cmd/8l/pass.c
@@ -444,11 +444,12 @@ dostkoff(void)
 
 		q = P;
 
-		if(!(p->from.scale & NOSPLIT)) {
+		if(!(p->from.scale & NOSPLIT) || (p->from.scale & WRAPPER)) {
 			p = appendp(p);
 			p = load_g_cx(p); // load g into CX
-			p = stacksplit(p, autoffset, &q); // emit split check
 		}
+		if(!(cursym->text->from.scale & NOSPLIT))
+			p = stacksplit(p, autoffset, &q); // emit split check
 
 		if(autoffset) {
 			p = appendp(p);
@@ -456,8 +457,6 @@ dostkoff(void)
 			p->from.type = D_CONST;
 			p->from.offset = autoffset;
 			p->spadj = autoffset;
-			if(q != P)
-				q->pcond = p;
 		} else {
 			// zero-byte stack adjustment.
 			// Insert a fake non-zero adjustment so that stkcheck can
@@ -469,8 +468,20 @@ dostkoff(void)
 			p->as = ANOP;
 			p->spadj = PtrSize;
 		}
+		if(q != P)
+			q->pcond = p;
 		deltasp = autoffset;
 		
+		if(cursym->text->from.scale & WRAPPER) {
+			// g->panicwrap += autoffset + PtrSize;
+			p = appendp(p);
+			p->as = AADDL;
+			p->from.type = D_CONST;
+			p->from.offset = autoffset + PtrSize;
+			p->to.type = D_INDIR+D_CX;
+			p->to.offset = 2*PtrSize;
+		}
+		
 		if(debug['Z'] && autoffset && !(cursym->text->from.scale&NOSPLIT)) {
 			// 8l -Z means zero the stack frame on entry.
 			// This slows down function calls but can help avoid
@@ -540,6 +551,19 @@ dostkoff(void)
 	
 			if(autoffset != deltasp)
 				diag("unbalanced PUSH/POP");
+
+			if(cursym->text->from.scale & WRAPPER) {
+				p = load_g_cx(p);
+				p = appendp(p);
+				// g->panicwrap -= autoffset + PtrSize;
+				p->as = ASUBL;
+				p->from.type = D_CONST;
+				p->from.offset = autoffset + PtrSize;
+				p->to.type = D_INDIR+D_CX;
+				p->to.offset = 2*PtrSize;
+				p = appendp(p);
+				p->as = ARET;
+			}
 	
 			if(autoffset) {
 				p->as = AADJSP;