Use a zio flag to prevent recursion of vdev_queue_io_done which can

cause stack overflow for IO's which return ZIO_PIPELINE_CONTINUE
from the zio_vdev_io_start stage and hence don't suspend and complete
in a different thread.

This prevents double fault panic on slow machines running ZFS on
GELI volumes which return EOPNOTSUPP directly to BIO_DELETE requests.

MFC after:	1 month
X-MFC-With:	r265152

2 files changed, 12 insertions, 0 deletions

diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio.h
index 2cce781..84187a8 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio.h
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio.h
@@ -196,6 +196,7 @@ enum zio_flag {
 	ZIO_FLAG_NOPWRITE	= 1 << 25,
 	ZIO_FLAG_REEXECUTED	= 1 << 26,
 	ZIO_FLAG_DELEGATED	= 1 << 27,
+	ZIO_FLAG_QUEUE_IO_DONE	= 1 << 28,
 };
 
 #define	ZIO_FLAG_MUSTSUCCEED		0
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_queue.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_queue.c
index 5747aa4..60fd4c3 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_queue.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_queue.c
@@ -796,14 +796,25 @@ vdev_queue_io_done(zio_t *zio)
 
 	vq->vq_io_complete_ts = gethrtime();
 
+	if (zio->io_flags & ZIO_FLAG_QUEUE_IO_DONE) {
+		/*
+		 * Executing from a previous vdev_queue_io_done so
+		 * to avoid recursion we just unlock and return.
+		 */
+		mutex_exit(&vq->vq_lock);
+		return;
+	}
+
 	while ((nio = vdev_queue_io_to_issue(vq)) != NULL) {
 		mutex_exit(&vq->vq_lock);
+		nio->io_flags |= ZIO_FLAG_QUEUE_IO_DONE;
 		if (nio->io_done == vdev_queue_agg_io_done) {
 			zio_nowait(nio);
 		} else {
 			zio_vdev_io_reissue(nio);
 			zio_execute(nio);
 		}
+		nio->io_flags &= ~ZIO_FLAG_QUEUE_IO_DONE;
 		mutex_enter(&vq->vq_lock);
 	}