MFC r274337,r274673,274681,r275515:

ZFS large block support.  The default recordsize remains at 128KB.

A new tunable/sysctl variable, vfs.zfs.max_recordsize is added to
allow adjusting the permitted maximum record size, or
zfs_max_recordsize, with a default of 1MB.  ZFS will not allow
setting recordsize greater than zfs_max_recordsize as a safety
belt, because larger recordsize means greater read and write
latency and more memory usage.

Please note that booting from datasets that have recordsize greater
than 128KB is not supported (but it's Okay to enable the feature on
the pool).

Limited safety belt is provided for mounted root filesystem but use
caution when using a larger value.

Illumos issue:
    5027 zfs large block support

1 files changed, 6 insertions, 0 deletions

diff --git a/sys/boot/zfs/zfsimpl.c b/sys/boot/zfs/zfsimpl.c
index 16b57c7..8d21c57 100644
--- a/sys/boot/zfs/zfsimpl.c
+++ b/sys/boot/zfs/zfsimpl.c
@@ -57,6 +57,7 @@ static const char *features_for_read[] = {
 	"com.delphix:hole_birth",
 	"com.delphix:extensible_dataset",
 	"com.delphix:embedded_data",
+	"org.open-zfs:large_blocks",
 	NULL
 };
 
@@ -1222,6 +1223,11 @@ dnode_read(const spa_t *spa, const dnode_phys_t *dnode, off_t offset, void *buf,
 	int nlevels = dnode->dn_nlevels;
 	int i, rc;
 
+	if (bsize > SPA_MAXBLOCKSIZE) {
+		printf("ZFS: I/O error - blocks larger than 128K are not supported\n");
+		return (EIO);
+	}
+
 	/*
 	 * Note: bsize may not be a power of two here so we need to do an
 	 * actual divide rather than a bitshift.