runtime: add mayAcquire annotation for trace.lock

Now that we've moved the trace locks to the leaf of the lock graph, we
can safely annotate that any trace event may acquire trace.lock even
if dynamically it turns out a particular event doesn't need to flush
and acquire this lock.

This reveals a new edge where we can trace while holding the mheap
lock, so we add this to the lock graph.

For #53789.
Updates #53979.

Change-Id: I13e2f6cd1b621cca4bed0cc13ef12e64d05c89a7
Reviewed-on: https://go-review.googlesource.com/c/go/+/418720
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>

src/runtime/mklockrank.go

@@ -159,6 +159,7 @@ mheap, mheapSpecial < globalAlloc;
 
 # Execution tracer events (with a P)
 hchan,
+  mheap,
   root,
   sched,
   traceStrings,

src/runtime/trace.go

@@ -883,6 +883,11 @@ func traceStackID(mp *m, buf []uintptr, skip int) uint64 {
 
 // traceAcquireBuffer returns trace buffer to use and, if necessary, locks it.
 func traceAcquireBuffer() (mp *m, pid int32, bufp *traceBufPtr) {
+	// Any time we acquire a buffer, we may end up flushing it,
+	// but flushes are rare. Record the lock edge even if it
+	// doesn't happen this time.
+	lockRankMayTraceFlush()
+
 	mp = acquirem()
 	if p := mp.p.ptr(); p != nil {
 		return mp, p.id, &p.tracebuf
@@ -899,6 +904,16 @@ func traceReleaseBuffer(pid int32) {
 	releasem(getg().m)
 }
 
+// lockRankMayTraceFlush records the lock ranking effects of a
+// potential call to traceFlush.
+func lockRankMayTraceFlush() {
+	owner := trace.lockOwner
+	dolock := owner == nil || owner != getg().m.curg
+	if dolock {
+		lockWithRankMayAcquire(&trace.lock, getLockRank(&trace.lock))
+	}
+}
+
 // traceFlush puts buf onto stack of full buffers and returns an empty buffer.
 //
 // This must run on the system stack because it acquires trace.lock.