runtime: don't lock mheap on user goroutine

This is bad for 2 reasons:
1. if the code under lock ever grows stack,
it will deadlock as stack growing acquires mheap lock.
2. It currently deadlocks with SetCPUProfileRate:
scavenger locks mheap, receives prof signal and tries to lock prof lock;
meanwhile SetCPUProfileRate locks prof lock and tries to grow stack
(presumably in runtime.unlock->futexwakeup). Boom.
Let's assume that it
Fixes #8407.

LGTM=rsc
R=golang-codereviews, rsc
CC=golang-codereviews, khr
https://golang.org/cl/112640043

diff --git a/src/pkg/runtime/mheap.c b/src/pkg/runtime/mheap.c
index 202a903ff82b2e1634b524731dba8ca4ffa43a76..7ae5a399b3564bc1e298f4457d46bfe15bfa29ef 100644 (file)
--- a/src/pkg/runtime/mheap.c
+++ b/src/pkg/runtime/mheap.c
@@ -642,7 +642,6 @@ static FuncVal forcegchelperv = {(void(*)(void))forcegchelper};
 void
 runtime·MHeap_Scavenger(void)
 {
-       MHeap *h;
        uint64 tick, forcegc, limit;
        int64 unixnow;
        int32 k;
@@ -662,15 +661,12 @@ runtime·MHeap_Scavenger(void)
        else
                tick = limit/2;
 
-       h = &runtime·mheap;
        for(k=0;; k++) {
                runtime·noteclear(&note);
                runtime·notetsleepg(&note, tick);
 
-               runtime·lock(h);
                unixnow = runtime·unixnanotime();
                if(unixnow - mstats.last_gc > forcegc) {
-                       runtime·unlock(h);
                        // The scavenger can not block other goroutines,
                        // otherwise deadlock detector can fire spuriously.
                        // GC blocks other goroutines via the runtime·worldsema.
@@ -680,13 +676,13 @@ runtime·MHeap_Scavenger(void)
                        runtime·notetsleepg(&note, -1);
                        if(runtime·debug.gctrace > 0)
                                runtime·printf("scvg%d: GC forced\n", k);
-                       runtime·lock(h);
                }
-               runtime·unlock(h);
+               g->m->locks++;  // ensure that we are on the same m while filling arguments
                g->m->scalararg[0] = k;
                g->m->scalararg[1] = runtime·nanotime();
                g->m->scalararg[2] = limit;
                runtime·mcall(scavenge_m);
+               g->m->locks--;
        }
 }