MFp4:

@118370	Correct typo.

@118371	Integrate changes from vendor.

@118491	Show backtrace on unexpected code paths.

@118494	Integrate changes from vendor.

@118504	Fix sendfile(2). I had two ways of fixing it:
	1. Fixing sendfile(2) itself to use VOP_GETPAGES() instead of
	   hacking around with vn_rdwr(UIO_NOCOPY), which was suggested
	   by ups.
	2. Modify ZFS behaviour to handle this special case.

	Although 1 is more correct, I've choosen 2, because hack from 1
	have a side-effect of beeing faster - it reads ahead MAXBSIZE
	bytes instead of reading page by page. This is not easy to implement
	with VOP_GETPAGES(), at least not for me in this very moment.

	Reported by:	Andrey V. Elsukov <bu7cher@yandex.ru>

@118525	Reorganize the code to reduce diff.

@118526	This code path is expected. It is simply when file is opened with
	O_FSYNC flag.

	Reported by:	kris
	Reported by:	Michal Suszko <dry@dry.pl>

5 files changed, 38 insertions, 25 deletions

diff --git a/sys/cddl/compat/opensolaris/kern/opensolaris_vfs.c b/sys/cddl/compat/opensolaris/kern/opensolaris_vfs.c
index 35404ca..ef184b4 100644
--- a/sys/cddl/compat/opensolaris/kern/opensolaris_vfs.c
+++ b/sys/cddl/compat/opensolaris/kern/opensolaris_vfs.c
@@ -206,7 +206,7 @@ domount(kthread_t *td, vnode_t *vp, const char *fstype, char *fspath,
 	mp->mnt_flag |= fsflags & (MNT_UPDATEMASK | MNT_FORCE | MNT_ROOTFS);
 	/*
 	 * Unprivileged user can trigger mounting a snapshot, but we don't want
-	 * him to unmount it, so we switch to privileged credential.
+	 * him to unmount it, so we switch to privileged credentials.
 	 */
 	crfree(mp->mnt_cred);
 	mp->mnt_cred = crdup(kcred);
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c
index f8d9465..c6a95a1 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c
@@ -499,10 +499,9 @@ spa_load(spa_t *spa, nvlist_t *config, spa_load_state_t state, int mosconfig)
 	/*
 	 * Try to open all vdevs, loading each label in the process.
 	 */
-	if (vdev_open(rvd) != 0) {
-		error = ENXIO;
+	error = vdev_open(rvd);
+	if (error != 0)
 		goto out;
-	}
 
 	/*
 	 * Validate the labels for all leaf vdevs.  We need to grab the config
@@ -513,10 +512,8 @@ spa_load(spa_t *spa, nvlist_t *config, spa_load_state_t state, int mosconfig)
 	error = vdev_validate(rvd);
 	spa_config_exit(spa, FTAG);
 
-	if (error != 0) {
-		error = EBADF;
+	if (error != 0)
 		goto out;
-	}
 
 	if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN) {
 		error = ENXIO;
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c
index 0fceb8d..d15f6e2 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c
@@ -931,7 +931,7 @@ vdev_validate(vdev_t *vd)
 
 	for (c = 0; c < vd->vdev_children; c++)
 		if (vdev_validate(vd->vdev_child[c]) != 0)
-			return (-1);
+			return (EBADF);
 
 	/*
 	 * If the device has already failed, or was marked offline, don't do
@@ -974,7 +974,7 @@ vdev_validate(vdev_t *vd)
 
 		if (spa->spa_load_state == SPA_LOAD_OPEN &&
 		    state != POOL_STATE_ACTIVE)
-			return (-1);
+			return (EBADF);
 	}
 
 	/*
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c
index 5f74eb8..7af6b20 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c
@@ -165,10 +165,8 @@ zfs_open(ap)
 	int flag = ap->a_mode;
 
 	/* Keep a count of the synchronous opens in the znode */
-	if (flag & FFSYNC) {
+	if (flag & FFSYNC)
 		atomic_inc_32(&zp->z_sync_cnt);
-		ZFS_LOG(0, "Unexpected code path, report to pjd@FreeBSD.org");
-	}
 
 	vnode_create_vobject(vp, zp->z_phys->zp_size, ap->a_td);
 	return (0);
@@ -188,10 +186,8 @@ zfs_close(ap)
 	int flag = ap->a_fflag;
 
 	/* Decrement the synchronous opens in the znode */
-	if (flag & FFSYNC) {
+	if (flag & FFSYNC)
 		atomic_dec_32(&zp->z_sync_cnt);
-		ZFS_LOG(0, "Unexpected code path, report to pjd@FreeBSD.org");
-	}
 
 	return (0);
 }
@@ -362,6 +358,7 @@ mappedread(vnode_t *vp, int nbytes, uio_t *uio)
 	vm_page_t m;
 	struct sf_buf *sf;
 	int64_t start, off;
+	caddr_t va;
 	int len = nbytes;
 	int error = 0;
 
@@ -378,8 +375,6 @@ mappedread(vnode_t *vp, int nbytes, uio_t *uio)
 again:
 		if ((m = vm_page_lookup(obj, OFF_TO_IDX(start))) != NULL &&
 		    vm_page_is_valid(m, (vm_offset_t)off, bytes)) {
-			caddr_t va;
-
 			if (vm_page_sleep_if_busy(m, FALSE, "zfsmrb"))
 				goto again;
 			vm_page_busy(m);
@@ -392,6 +387,21 @@ again:
 			sched_unpin();
 			VM_OBJECT_LOCK(obj);
 			vm_page_wakeup(m);
+		} else if (m != NULL && uio->uio_segflg == UIO_NOCOPY) {
+			if (vm_page_sleep_if_busy(m, FALSE, "zfsmrb"))
+				goto again;
+			vm_page_busy(m);
+			VM_OBJECT_UNLOCK(obj);
+			sched_pin();
+			sf = sf_buf_alloc(m, SFB_CPUPRIVATE);
+			va = (caddr_t)sf_buf_kva(sf);
+			error = dmu_read(os, zp->z_id, start + off, bytes,
+			    (void *)(va + off));
+			sf_buf_free(sf);
+			sched_unpin();
+			VM_OBJECT_LOCK(obj);
+			vm_page_wakeup(m);
+			uio->uio_resid -= bytes;
 		} else {
 			VM_OBJECT_UNLOCK(obj);
 			error = dmu_read_uio(os, zp->z_id, uio, bytes);
@@ -2461,7 +2471,7 @@ top:
 		if (tvp)
 			vrele(tvp);
 		else {
-			ZFS_LOG(0, "Unexpected code path, report to pjd@FreeBSD.org");
+			ZFS_LOG(0x100, "Unexpected code path, report to pjd@FreeBSD.org");
 			tvp = ZTOV(tzp);
 			vn_lock(tvp, LK_EXCLUSIVE | LK_RETRY, curthread);
 		}
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c
index 6bc4a36..99c0dd3 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c
@@ -79,12 +79,13 @@ zio_sync_pass_t zio_sync_pass = {
 	1,	/* zp_rewrite */
 };
 
-#ifdef ZIO_USE_UMA
 /*
  * ==========================================================================
  * I/O kmem caches
  * ==========================================================================
  */
+kmem_cache_t *zio_cache;
+#ifdef ZIO_USE_UMA
 kmem_cache_t *zio_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT];
 kmem_cache_t *zio_data_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT];
 #endif
@@ -107,6 +108,9 @@ zio_init(void)
 #endif
 #endif
 
+	zio_cache = kmem_cache_create("zio_cache", sizeof (zio_t), 0,
+	    NULL, NULL, NULL, NULL, NULL, 0);
+
 #ifdef ZIO_USE_UMA
 	/*
 	 * For small buffers, we want a cache for each multiple of
@@ -183,6 +187,8 @@ zio_fini(void)
 	}
 #endif
 
+	kmem_cache_destroy(zio_cache);
+
 	zio_inject_fini();
 }
 
@@ -329,7 +335,8 @@ zio_create(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp,
 	ASSERT3U(size, <=, SPA_MAXBLOCKSIZE);
 	ASSERT(P2PHASE(size, SPA_MINBLOCKSIZE) == 0);
 
-	zio = kmem_zalloc(sizeof (zio_t), KM_SLEEP);
+	zio = kmem_cache_alloc(zio_cache, KM_SLEEP);
+	bzero(zio, sizeof (zio_t));
 	zio->io_parent = pio;
 	zio->io_spa = spa;
 	zio->io_txg = txg;
@@ -777,7 +784,7 @@ zio_wait(zio_t *zio)
 	error = zio->io_error;
 	cv_destroy(&zio->io_cv);
 	mutex_destroy(&zio->io_lock);
-	kmem_free(zio, sizeof (zio_t));
+	kmem_cache_free(zio_cache, zio);
 
 	return (error);
 }
@@ -957,9 +964,8 @@ zio_done(zio_t *zio)
 	}
 
 	/*
-	 * Note: this I/O is now done, and will shortly be
-	 * kmem_free()'d, so there is no need to clear this (or any
-	 * other) flag.
+	 * Note: this I/O is now done, and will shortly be freed, so there is no
+	 * need to clear this (or any other) flag.
 	 */
 	if (zio->io_flags & ZIO_FLAG_CONFIG_GRABBED)
 		spa_config_exit(spa, zio);
@@ -971,7 +977,7 @@ zio_done(zio_t *zio)
 		cv_broadcast(&zio->io_cv);
 		mutex_exit(&zio->io_lock);
 	} else {
-		kmem_free(zio, sizeof (zio_t));
+		kmem_cache_free(zio_cache, zio);
 	}
 }