Un-staticize runningbufwakeup() and staticize updateproc.

Add a new private thread flag to indicate that the thread should
not sleep if runningbufspace is too large.

Set this flag on the bufdaemon and syncer threads so that they skip
the waitrunningbufspace() call in bufwrite() rather than than
checking the proc pointer vs. the known proc pointers for these two
threads.  A way of preventing these threads from being starved for
I/O but still placing limits on their outstanding I/O would be
desirable.

Set this flag in ffs_copyonwrite() to prevent bufwrite() calls from
blocking on the runningbufspace check while holding snaplk.  This
prevents snaplk from being held for an arbitrarily long period of
time if runningbufspace is high and greatly reduces the contention
for snaplk.  The disadvantage is that ffs_copyonwrite() can start
a large amount of I/O if there are a large number of snapshots,
which could cause a deadlock in other parts of the code.

Call runningbufwakeup() in ffs_copyonwrite() to decrement runningbufspace
before attempting to grab snaplk so that I/O requests waiting on
snaplk are not counted in runningbufspace as being in-progress.
Increment runningbufspace again before actually launching the
original I/O request.

Prior to the above two changes, the system could deadlock if enough
I/O requests were blocked by snaplk to prevent runningbufspace from
falling below lorunningspace and one of the bawrite() calls in
ffs_copyonwrite() blocked in waitrunningbufspace() while holding
snaplk.

See <http://www.holm.cc/stress/log/cons143.html>

1 files changed, 16 insertions, 4 deletions

diff --git a/sys/ufs/ffs/ffs_snapshot.c b/sys/ufs/ffs/ffs_snapshot.c
index c134461..3e7e231 100644
--- a/sys/ufs/ffs/ffs_snapshot.c
+++ b/sys/ufs/ffs/ffs_snapshot.c
@@ -1998,6 +1998,12 @@ ffs_copyonwrite(devvp, bp)
 		return (0);
 	}
 	/*
+	 * Since I/O on bp isn't yet in progress and it may be blocked
+	 * for a long time waiting on snaplk, back it out of
+	 * runningbufspace, possibly waking other threads waiting for space.
+	 */
+	runningbufwakeup(bp);
+	/*
 	 * Not in the precomputed list, so check the snapshots.
 	 */
 retry:
@@ -2028,7 +2034,7 @@ retry:
 				goto retry;
 			}
 			snapshot_locked = 1;
-			td->td_pflags |= TDP_COWINPROGRESS;
+			td->td_pflags |= TDP_COWINPROGRESS | TDP_NORUNNINGBUF;
 			error = UFS_BALLOC(vp, lblktosize(fs, (off_t)lbn),
 			   fs->fs_bsize, KERNCRED, BA_METAONLY, &ibp);
 			td->td_pflags &= ~TDP_COWINPROGRESS;
@@ -2065,7 +2071,7 @@ retry:
 			goto retry;
 		}
 		snapshot_locked = 1;
-		td->td_pflags |= TDP_COWINPROGRESS;
+		td->td_pflags |= TDP_COWINPROGRESS | TDP_NORUNNINGBUF;
 		error = UFS_BALLOC(vp, lblktosize(fs, (off_t)lbn),
 		    fs->fs_bsize, KERNCRED, 0, &cbp);
 		td->td_pflags &= ~TDP_COWINPROGRESS;
@@ -2120,10 +2126,16 @@ retry:
 		if (dopersistence && VTOI(vp)->i_effnlink > 0)
 			(void) ffs_syncvnode(vp, MNT_WAIT);
 	}
-	if (snapshot_locked)
+	if (snapshot_locked) {
 		lockmgr(vp->v_vnlock, LK_RELEASE, NULL, td);
-	else
+		td->td_pflags &= ~TDP_NORUNNINGBUF;
+	} else
 		VI_UNLOCK(devvp);
+	/*
+	 * I/O on bp will now be started, so count it in runningbufspace.
+	 */
+	if (bp->b_runningbufspace)
+		atomic_add_int(&runningbufspace, bp->b_runningbufspace);
 	return (error);
 }