Revert "Illumos openzfs#5244 zio pipeline callers should explicitly i…

…nvoke next stage openzfs#2828"

This reverts commit 9291279.

conflicts with:

commit 92119cc
Author: Brian Behlendorf <behlendorf1@llnl.gov>
Date:   Sun Jul 13 14:35:19 2014 -0400

    Mark IO pipeline with PF_FSTRANS

    In order to avoid deadlocking in the IO pipeline it is critical that
    pageout be avoided during direct memory reclaim.  This ensures that
    the pipeline threads can always make forward progress and never end
    up blocking on a DMU transaction.  For this very reason Linux now
    provides the PF_FSTRANS flag which may be set in the process context.

    Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>

not sure where that commit came from, it's not in master (?!) but commits
where always fetched from master /sadpanda

include/sys/vdev_impl.h

@@ -60,7 +60,7 @@ typedef int	vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size,
     uint64_t *ashift);
 typedef void	vdev_close_func_t(vdev_t *vd);
 typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize);
-typedef void	vdev_io_start_func_t(zio_t *zio);
+typedef int	vdev_io_start_func_t(zio_t *zio);
 typedef void	vdev_io_done_func_t(zio_t *zio);
 typedef void	vdev_state_change_func_t(vdev_t *vd, int, int);
 typedef void	vdev_hold_func_t(vdev_t *vd);

include/sys/zio.h

@@ -154,6 +154,9 @@ typedef enum zio_priority {
 	ZIO_PRIORITY_NOW		/* non-queued i/os (e.g. free) */
 } zio_priority_t;
 
+#define	ZIO_PIPELINE_CONTINUE		0x100
+#define	ZIO_PIPELINE_STOP		0x101
+
 enum zio_flag {
 	/*
 	 * Flags inherited by gang, ddt, and vdev children,

lib/libzpool/taskq.c

@@ -25,7 +25,6 @@
 /*
  * Copyright 2011 Nexenta Systems, Inc. All rights reserved.
  * Copyright 2012 Garrett D'Amore <garrett@damore.org>.  All rights reserved.
- * Copyright (c) 2014 by Delphix. All rights reserved.
  */
 
 #include <sys/zfs_context.h>
@@ -34,10 +33,8 @@ int taskq_now;
 taskq_t *system_taskq;
 
 #define	TASKQ_ACTIVE	0x00010000
-#define	TASKQ_NAMELEN	31
 
 struct taskq {
-	char		tq_name[TASKQ_NAMELEN + 1];
 	kmutex_t	tq_lock;
 	krwlock_t	tq_threadlock;
 	kcondvar_t	tq_dispatch_cv;
@@ -283,7 +280,6 @@ taskq_create(const char *name, int nthreads, pri_t pri,
 	cv_init(&tq->tq_dispatch_cv, NULL, CV_DEFAULT, NULL);
 	cv_init(&tq->tq_wait_cv, NULL, CV_DEFAULT, NULL);
 	cv_init(&tq->tq_maxalloc_cv, NULL, CV_DEFAULT, NULL);
-	(void) strncpy(tq->tq_name, name, TASKQ_NAMELEN + 1);
 	tq->tq_flags = flags | TASKQ_ACTIVE;
 	tq->tq_active = nthreads;
 	tq->tq_nthreads = nthreads;

module/zfs/vdev_disk.c

@@ -671,7 +671,7 @@ vdev_disk_io_flush(struct block_device *bdev, zio_t *zio)
 	return (0);
 }
 
-static void
+static int
 vdev_disk_io_start(zio_t *zio)
 {
 	vdev_t *v = zio->io_vd;
@@ -683,9 +683,7 @@ vdev_disk_io_start(zio_t *zio)
 
 		if (!vdev_readable(v)) {
 			zio->io_error = SET_ERROR(ENXIO);
-			zio_interrupt(zio);
-			return;
-
+			return (ZIO_PIPELINE_CONTINUE);
 		}
 
 		switch (zio->io_cmd) {
@@ -701,7 +699,7 @@ vdev_disk_io_start(zio_t *zio)
 
 			error = vdev_disk_io_flush(vd->vd_bdev, zio);
 			if (error == 0)
-				return;
+				return (ZIO_PIPELINE_STOP);
 
 			zio->io_error = error;
 			if (error == ENOTSUP)
@@ -713,8 +711,8 @@ vdev_disk_io_start(zio_t *zio)
 			zio->io_error = SET_ERROR(ENOTSUP);
 		}
 
-		zio_execute(zio);
-		return;
+		return (ZIO_PIPELINE_CONTINUE);
+
 	case ZIO_TYPE_WRITE:
 		flags = WRITE;
 		break;
@@ -725,17 +723,17 @@ vdev_disk_io_start(zio_t *zio)
 
 	default:
 		zio->io_error = SET_ERROR(ENOTSUP);
-		zio_interrupt(zio);
-		return;
+		return (ZIO_PIPELINE_CONTINUE);
 	}
 
 	error = __vdev_disk_physio(vd->vd_bdev, zio, NULL,
 	    zio->io_size, zio->io_offset, flags);
 	if (error) {
 		zio->io_error = error;
-		zio_interrupt(zio);
-		return;
+		return (ZIO_PIPELINE_CONTINUE);
 	}
+
+	return (ZIO_PIPELINE_STOP);
 }
 
 static void

module/zfs/vdev_file.c

@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  */
 
 #include <sys/zfs_context.h>
@@ -183,7 +183,7 @@ vdev_file_io_fsync(void *arg)
 	zio_interrupt(zio);
 }
 
-static void
+static int
 vdev_file_io_start(zio_t *zio)
 {
 	vdev_t *vd = zio->io_vd;
@@ -193,8 +193,7 @@ vdev_file_io_start(zio_t *zio)
 		/* XXPOLICY */
 		if (!vdev_readable(vd)) {
 			zio->io_error = SET_ERROR(ENXIO);
-			zio_interrupt(zio);
-			return;
+			return (ZIO_PIPELINE_CONTINUE);
 		}
 
 		switch (zio->io_cmd) {
@@ -223,12 +222,13 @@ vdev_file_io_start(zio_t *zio)
 			zio->io_error = SET_ERROR(ENOTSUP);
 		}
 
-		zio_execute(zio);
-		return;
+		return (ZIO_PIPELINE_CONTINUE);
 	}
 
 	VERIFY3U(taskq_dispatch(vdev_file_taskq, vdev_file_io_strategy, zio,
 	    TQ_SLEEP), !=, 0);
+
+	return (ZIO_PIPELINE_STOP);
 }
 
 /* ARGSUSED */

module/zfs/vdev_mirror.c

@@ -24,7 +24,7 @@
  */
 
 /*
- * Copyright (c) 2012, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  */
 
 #include <sys/zfs_context.h>
@@ -326,7 +326,7 @@ vdev_mirror_child_select(zio_t *zio)
 	return (-1);
 }
 
-static void
+static int
 vdev_mirror_io_start(zio_t *zio)
 {
 	mirror_map_t *mm;
@@ -357,8 +357,7 @@ vdev_mirror_io_start(zio_t *zio)
 				    zio->io_type, zio->io_priority, 0,
 				    vdev_mirror_scrub_done, mc));
 			}
-			zio_execute(zio);
-			return;
+			return (ZIO_PIPELINE_CONTINUE);
 		}
 		/*
 		 * For normal reads just pick one child.
@@ -384,7 +383,7 @@ vdev_mirror_io_start(zio_t *zio)
 		c++;
 	}
 
-	zio_execute(zio);
+	return (ZIO_PIPELINE_CONTINUE);
 }
 
 static int

module/zfs/vdev_missing.c

@@ -24,7 +24,7 @@
  */
 
 /*
- * Copyright (c) 2012, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  */
 
 /*
@@ -66,11 +66,11 @@ vdev_missing_close(vdev_t *vd)
 }
 
 /* ARGSUSED */
-static void
+static int
 vdev_missing_io_start(zio_t *zio)
 {
 	zio->io_error = SET_ERROR(ENOTSUP);
-	zio_execute(zio);
+	return (ZIO_PIPELINE_CONTINUE);
 }
 
 /* ARGSUSED */

module/zfs/vdev_raidz.c

@@ -21,7 +21,7 @@
 
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2013 by Delphix. All rights reserved.
  */
 
 #include <sys/zfs_context.h>
@@ -1722,7 +1722,7 @@ vdev_raidz_child_done(zio_t *zio)
  *      vdevs have had errors, then create zio read operations to the parity
  *      columns' VDevs as well.
  */
-static void
+static int
 vdev_raidz_io_start(zio_t *zio)
 {
 	vdev_t *vd = zio->io_vd;
@@ -1766,8 +1766,7 @@ vdev_raidz_io_start(zio_t *zio)
 			    ZIO_FLAG_NODATA | ZIO_FLAG_OPTIONAL, NULL, NULL));
 		}
 
-		zio_execute(zio);
-		return;
+		return (ZIO_PIPELINE_CONTINUE);
 	}
 
 	ASSERT(zio->io_type == ZIO_TYPE_READ);
@@ -1807,7 +1806,7 @@ vdev_raidz_io_start(zio_t *zio)
 		}
 	}
 
-	zio_execute(zio);
+	return (ZIO_PIPELINE_CONTINUE);
 }
 
 

module/zfs/zio.c

@@ -59,9 +59,6 @@ kmem_cache_t *zio_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT];
 kmem_cache_t *zio_data_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT];
 int zio_delay_max = ZIO_DELAY_MAX;
 
-#define	ZIO_PIPELINE_CONTINUE		0x100
-#define	ZIO_PIPELINE_STOP		0x101
-
 /*
  * The following actions directly effect the spa's sync-to-convergence logic.
  * The values below define the sync pass when we start performing the action.
@@ -2611,18 +2608,6 @@ zio_free_zil(spa_t *spa, uint64_t txg, blkptr_t *bp)
  * Read and write to physical devices
  * ==========================================================================
  */
-
-
-/*
- * Issue an I/O to the underlying vdev. Typically the issue pipeline
- * stops after this stage and will resume upon I/O completion.
- * However, there are instances where the vdev layer may need to
- * continue the pipeline when an I/O was not issued. Since the I/O
- * that was sent to the vdev layer might be different than the one
- * currently active in the pipeline (see vdev_queue_io()), we explicitly
- * force the underlying vdev layers to call either zio_execute() or
- * zio_interrupt() to ensure that the pipeline continues with the correct I/O.
- */
 static int
 zio_vdev_io_start(zio_t *zio)
 {
@@ -2640,16 +2625,15 @@ zio_vdev_io_start(zio_t *zio)
 		/*
 		 * The mirror_ops handle multiple DVAs in a single BP.
 		 */
-		vdev_mirror_ops.vdev_op_io_start(zio);
-		return (ZIO_PIPELINE_STOP);
+		return (vdev_mirror_ops.vdev_op_io_start(zio));
 	}
 
 	/*
 	 * We keep track of time-sensitive I/Os so that the scan thread
 	 * can quickly react to certain workloads.  In particular, we care
 	 * about non-scrubbing, top-level reads and writes with the following
 	 * characteristics:
-	 *	- synchronous writes of user data to non-slog devices
+	 * 	- synchronous writes of user data to non-slog devices
 	 *	- any reads of user data
 	 * When these conditions are met, adjust the timestamp of spa_last_io
 	 * which allows the scan thread to adjust its workload accordingly.
@@ -2739,8 +2723,7 @@ zio_vdev_io_start(zio_t *zio)
 		}
 	}
 
-	vd->vdev_ops->vdev_op_io_start(zio);
-	return (ZIO_PIPELINE_STOP);
+	return (vd->vdev_ops->vdev_op_io_start(zio));
 }
 
 static int