[block] Replace gPXE block-device API with an iPXE asynchronous interface

The block device interface used in gPXE predates the invention of even the old gPXE data-transfer interface, let alone the current iPXE generic asynchronous interface mechanism. Bring this old code up to date, with the following benefits: o Block device commands can be cancelled by the requestor. The INT 13 layer uses this to provide a global timeout on all INT 13 calls, with the result that an unexpected passive failure mode (such as an iSCSI target ACKing the request but never sending a response) will lead to a timeout that gets reported back to the INT 13 user, rather than simply freezing the system. o INT 13,00 (reset drive) is now able to reset the underlying block device. INT 13 users, such as DOS, that use INT 13,00 as a method for error recovery now have a chance of recovering. o All block device commands are tagged, with a numerical tag that will show up in debugging output and in packet captures; this will allow easier interpretation of bug reports that include both sources of information. o The extremely ugly hacks used to generate the boot firmware tables have been eradicated and replaced with a generic acpi_describe() method (exploiting the ability of iPXE interfaces to pass through methods to an underlying interface). The ACPI tables are now built in a shared data block within .bss16, rather than each requiring dedicated space in .data16. o The architecture-independent concept of a SAN device has been exposed to the iPXE core through the sanboot API, which provides calls to hook, unhook, boot, and describe SAN devices. This allows for much more flexible usage patterns (such as hooking an empty SAN device and then running an OS installer via TFTP). Signed-off-by: Michael Brown <mcb30@ipxe.org>
2026-02-28 03:11:18 +03:00 · 2010-09-03 16:11:51 +01:00
parent 46c7f99c66
commit 220495f8bf
47 changed files with 4904 additions and 3127 deletions
--- a/src/net/aoe.c
+++ b/src/net/aoe.c
--- a/src/net/infiniband/ib_srp.c
+++ b/src/net/infiniband/ib_srp.c
@@ -32,7 +32,11 @@ FILE_LICENCE ( BSD2 );

 #include <stdlib.h>
 #include <errno.h>
+#include <ipxe/interface.h>
+#include <ipxe/uri.h>
+#include <ipxe/open.h>
 #include <ipxe/base16.h>
+#include <ipxe/acpi.h>
 #include <ipxe/srp.h>
 #include <ipxe/infiniband.h>
 #include <ipxe/ib_cmrc.h>
@@ -56,6 +60,192 @@ FILE_LICENCE ( BSD2 );
 #define EINFO_EINVAL_RP_TOO_SHORT __einfo_uniqify \
 	( EINFO_EINVAL, 0x04, "Root path too short" )

+/******************************************************************************
+ *
+ * IB SRP devices
+ *
+ ******************************************************************************
+ */
+
+/** An Infiniband SRP device */
+struct ib_srp_device {
+	/** Reference count */
+	struct refcnt refcnt;
+
+	/** SRP transport interface */
+	struct interface srp;
+	/** CMRC interface */
+	struct interface cmrc;
+
+	/** Infiniband device */
+	struct ib_device *ibdev;
+
+	/** Destination GID (for boot firmware table) */
+	struct ib_gid dgid;
+	/** Service ID (for boot firmware table) */
+	struct ib_gid_half service_id;
+};
+
+/**
+ * Free IB SRP device
+ *
+ * @v refcnt		Reference count
+ */
+static void ib_srp_free ( struct refcnt *refcnt ) {
+	struct ib_srp_device *ib_srp =
+		container_of ( refcnt, struct ib_srp_device, refcnt );
+
+	ibdev_put ( ib_srp->ibdev );
+	free ( ib_srp );
+}
+
+/**
+ * Close IB SRP device
+ *
+ * @v ib_srp		IB SRP device
+ * @v rc		Reason for close
+ */
+static void ib_srp_close ( struct ib_srp_device *ib_srp, int rc ) {
+
+	/* Shut down interfaces */
+	intf_shutdown ( &ib_srp->cmrc, rc );
+	intf_shutdown ( &ib_srp->srp, rc );
+}
+
+/**
+ * Describe IB SRP device in an ACPI table
+ *
+ * @v srpdev		SRP device
+ * @v acpi		ACPI table
+ * @v len		Length of ACPI table
+ * @ret rc		Return status code
+ */
+static int ib_srp_describe ( struct ib_srp_device *ib_srp,
+			     struct acpi_description_header *acpi,
+			     size_t len ) {
+	struct ib_device *ibdev = ib_srp->ibdev;
+	struct sbft_table *sbft =
+		container_of ( acpi, struct sbft_table, acpi );
+	struct sbft_ib_subtable *ib_sbft;
+	size_t used;
+
+	/* Sanity check */
+	if ( acpi->signature != SBFT_SIG )
+		return -EINVAL;
+
+	/* Append IB subtable to existing table */
+	used = le32_to_cpu ( sbft->acpi.length );
+	sbft->ib_offset = cpu_to_le16 ( used );
+	ib_sbft = ( ( ( void * ) sbft ) + used );
+	used += sizeof ( *ib_sbft );
+	if ( used > len )
+		return -ENOBUFS;
+	sbft->acpi.length = cpu_to_le32 ( used );
+
+	/* Populate subtable */
+	memcpy ( &ib_sbft->sgid, &ibdev->gid, sizeof ( ib_sbft->sgid ) );
+	memcpy ( &ib_sbft->dgid, &ib_srp->dgid, sizeof ( ib_sbft->dgid ) );
+	memcpy ( &ib_sbft->service_id, &ib_srp->service_id,
+		 sizeof ( ib_sbft->service_id ) );
+	ib_sbft->pkey = cpu_to_le16 ( ibdev->pkey );
+
+	return 0;
+}
+
+/** IB SRP CMRC interface operations */
+static struct interface_operation ib_srp_cmrc_op[] = {
+	INTF_OP ( intf_close, struct ib_srp_device *, ib_srp_close ),
+};
+
+/** IB SRP CMRC interface descriptor */
+static struct interface_descriptor ib_srp_cmrc_desc =
+	INTF_DESC_PASSTHRU ( struct ib_srp_device, cmrc, ib_srp_cmrc_op, srp );
+
+/** IB SRP SRP interface operations */
+static struct interface_operation ib_srp_srp_op[] = {
+	INTF_OP ( acpi_describe, struct ib_srp_device *, ib_srp_describe ),
+	INTF_OP ( intf_close, struct ib_srp_device *, ib_srp_close ),
+};
+
+/** IB SRP SRP interface descriptor */
+static struct interface_descriptor ib_srp_srp_desc =
+	INTF_DESC_PASSTHRU ( struct ib_srp_device, srp, ib_srp_srp_op, cmrc );
+
+/**
+ * Open IB SRP device
+ *
+ * @v block		Block control interface
+ * @v ibdev		Infiniband device
+ * @v dgid		Destination GID
+ * @v service_id	Service ID
+ * @v initiator		Initiator port ID
+ * @v target		Target port ID
+ * @v lun		SCSI LUN
+ * @ret rc		Return status code
+ */
+static int ib_srp_open ( struct interface *block, struct ib_device *ibdev,
+			 struct ib_gid *dgid, struct ib_gid_half *service_id,
+			 union srp_port_id *initiator,
+			 union srp_port_id *target, struct scsi_lun *lun ) {
+	struct ib_srp_device *ib_srp;
+	int rc;
+
+	/* Allocate and initialise structure */
+	ib_srp = zalloc ( sizeof ( *ib_srp ) );
+	if ( ! ib_srp ) {
+		rc = -ENOMEM;
+		goto err_zalloc;
+	}
+	ref_init ( &ib_srp->refcnt, ib_srp_free );
+	intf_init ( &ib_srp->srp, &ib_srp_srp_desc, &ib_srp->refcnt );
+	intf_init ( &ib_srp->cmrc, &ib_srp_cmrc_desc, &ib_srp->refcnt );
+	ib_srp->ibdev = ibdev_get ( ibdev );
+	DBGC ( ib_srp, "IBSRP %p created for %08x%08x%08x%08x:%08x%08x\n",
+	       ib_srp, ntohl ( dgid->u.dwords[0] ),
+	       ntohl ( dgid->u.dwords[1] ), ntohl ( dgid->u.dwords[2] ),
+	       ntohl ( dgid->u.dwords[3] ), ntohl ( service_id->u.dwords[0] ),
+	       ntohl ( service_id->u.dwords[1] ) );
+
+	/* Preserve parameters required for boot firmware table */
+	memcpy ( &ib_srp->dgid, dgid, sizeof ( ib_srp->dgid ) );
+	memcpy ( &ib_srp->service_id, service_id,
+		 sizeof ( ib_srp->service_id ) );
+
+	/* Open CMRC socket */
+	if ( ( rc = ib_cmrc_open ( &ib_srp->cmrc, ibdev, dgid,
+				   service_id ) ) != 0 ) {
+		DBGC ( ib_srp, "IBSRP %p could not open CMRC socket: %s\n",
+		       ib_srp, strerror ( rc ) );
+		goto err_cmrc_open;
+	}
+
+	/* Attach SRP device to parent interface */
+	if ( ( rc = srp_open ( block, &ib_srp->srp, initiator, target,
+			       ibdev->rdma_key, lun ) ) != 0 ) {
+		DBGC ( ib_srp, "IBSRP %p could not create SRP device: %s\n",
+		       ib_srp, strerror ( rc ) );
+		goto err_srp_open;
+	}
+
+	/* Mortalise self and return */
+	ref_put ( &ib_srp->refcnt );
+	return 0;
+
+ err_srp_open:
+ err_cmrc_open:
+	ib_srp_close ( ib_srp, rc );
+	ref_put ( &ib_srp->refcnt );
+ err_zalloc:
+	return rc;
+}
+
+/******************************************************************************
+ *
+ * IB SRP URIs
+ *
+ ******************************************************************************
+ */
+
 /** IB SRP parse flags */
 enum ib_srp_parse_flags {
 	IB_SRP_PARSE_REQUIRED = 0x0000,
@@ -65,12 +255,20 @@ enum ib_srp_parse_flags {

 /** IB SRP root path parameters */
 struct ib_srp_root_path {
+	/** Source GID */
+	struct ib_gid sgid;
+	/** Initiator port ID */
+	union ib_srp_initiator_port_id initiator;
+	/** Destination GID */
+	struct ib_gid dgid;
+	/** Partition key */
+	uint16_t pkey;
+	/** Service ID */
+	struct ib_gid_half service_id;
 	/** SCSI LUN */
-	struct scsi_lun *lun;
-	/** SRP port IDs */
-	struct srp_port_ids *port_ids;
-	/** IB SRP parameters */
-	struct ib_srp_parameters *ib;
+	struct scsi_lun lun;
+	/** Target port ID */
+	union ib_srp_target_port_id target;
 };

 /**
@@ -99,7 +297,6 @@ static int ib_srp_parse_byte_string ( const char *rp_comp, uint8_t *bytes,
 	decoded_size = base16_decode ( rp_comp, bytes );
 	if ( decoded_size < 0 )
 		return decoded_size;
-	assert ( decoded_size == size );

 	return 0;
 }
@@ -125,19 +322,6 @@ static int ib_srp_parse_integer ( const char *rp_comp, int default_value ) {
 	return value;
 }

-/**
- * Parse IB SRP root path literal component
- *
- * @v rp_comp		Root path component string
- * @v rp		IB SRP root path
- * @ret rc		Return status code
- */
-static int ib_srp_parse_literal ( const char *rp_comp __unused,
-				  struct ib_srp_root_path *rp __unused ) {
-	/* Ignore */
-	return 0;
-}
-
 /**
 * Parse IB SRP root path source GID
 *
@@ -151,10 +335,10 @@ static int ib_srp_parse_sgid ( const char *rp_comp,

 	/* Default to the GID of the last opened Infiniband device */
 	if ( ( ibdev = last_opened_ibdev() ) != NULL )
-		memcpy ( &rp->ib->sgid, &ibdev->gid, sizeof ( rp->ib->sgid ) );
+		memcpy ( &rp->sgid, &ibdev->gid, sizeof ( rp->sgid ) );

-	return ib_srp_parse_byte_string ( rp_comp, rp->ib->sgid.u.bytes,
-					  ( sizeof ( rp->ib->sgid ) |
+	return ib_srp_parse_byte_string ( rp_comp, rp->sgid.u.bytes,
+					  ( sizeof ( rp->sgid ) |
 					    IB_SRP_PARSE_OPTIONAL ) );
 }

@@ -167,11 +351,10 @@ static int ib_srp_parse_sgid ( const char *rp_comp,
 */
 static int ib_srp_parse_initiator_id_ext ( const char *rp_comp,
 					   struct ib_srp_root_path *rp ) {
-	struct ib_srp_initiator_port_id *port_id =
-		ib_srp_initiator_port_id ( rp->port_ids );
+	union ib_srp_initiator_port_id *port_id = &rp->initiator;

-	return ib_srp_parse_byte_string ( rp_comp, port_id->id_ext.u.bytes,
-					  ( sizeof ( port_id->id_ext ) |
+	return ib_srp_parse_byte_string ( rp_comp, port_id->ib.id_ext.u.bytes,
+					  ( sizeof ( port_id->ib.id_ext ) |
 					    IB_SRP_PARSE_OPTIONAL ) );
 }

@@ -184,15 +367,14 @@ static int ib_srp_parse_initiator_id_ext ( const char *rp_comp,
 */
 static int ib_srp_parse_initiator_hca_guid ( const char *rp_comp,
 					     struct ib_srp_root_path *rp ) {
-	struct ib_srp_initiator_port_id *port_id =
-		ib_srp_initiator_port_id ( rp->port_ids );
+	union ib_srp_initiator_port_id *port_id = &rp->initiator;

 	/* Default to the GUID portion of the source GID */
-	memcpy ( &port_id->hca_guid, &rp->ib->sgid.u.half[1],
-		 sizeof ( port_id->hca_guid ) );
+	memcpy ( &port_id->ib.hca_guid, &rp->sgid.u.half[1],
+		 sizeof ( port_id->ib.hca_guid ) );

-	return ib_srp_parse_byte_string ( rp_comp, port_id->hca_guid.u.bytes,
-					  ( sizeof ( port_id->hca_guid ) |
+	return ib_srp_parse_byte_string ( rp_comp, port_id->ib.hca_guid.u.bytes,
+					  ( sizeof ( port_id->ib.hca_guid ) |
 					    IB_SRP_PARSE_OPTIONAL ) );
 }

@@ -205,8 +387,8 @@ static int ib_srp_parse_initiator_hca_guid ( const char *rp_comp,
 */
 static int ib_srp_parse_dgid ( const char *rp_comp,
 			       struct ib_srp_root_path *rp ) {
-	return ib_srp_parse_byte_string ( rp_comp, rp->ib->dgid.u.bytes,
-					  ( sizeof ( rp->ib->dgid ) |
+	return ib_srp_parse_byte_string ( rp_comp, rp->dgid.u.bytes,
+					  ( sizeof ( rp->dgid ) |
 					    IB_SRP_PARSE_REQUIRED ) );
 }

@@ -223,7 +405,7 @@ static int ib_srp_parse_pkey ( const char *rp_comp,

 	if ( ( pkey = ib_srp_parse_integer ( rp_comp, IB_PKEY_DEFAULT ) ) < 0 )
 		return pkey;
-	rp->ib->pkey = pkey;
+	rp->pkey = pkey;
 	return 0;
 }

@@ -236,8 +418,8 @@ static int ib_srp_parse_pkey ( const char *rp_comp,
 */
 static int ib_srp_parse_service_id ( const char *rp_comp,
 				     struct ib_srp_root_path *rp ) {
-	return ib_srp_parse_byte_string ( rp_comp, rp->ib->service_id.u.bytes,
-					  ( sizeof ( rp->ib->service_id ) |
+	return ib_srp_parse_byte_string ( rp_comp, rp->service_id.u.bytes,
+					  ( sizeof ( rp->service_id ) |
 					    IB_SRP_PARSE_REQUIRED ) );
 }

@@ -250,7 +432,7 @@ static int ib_srp_parse_service_id ( const char *rp_comp,
 */
 static int ib_srp_parse_lun ( const char *rp_comp,
 			      struct ib_srp_root_path *rp ) {
-	return scsi_parse_lun ( rp_comp, rp->lun );
+	return scsi_parse_lun ( rp_comp, &rp->lun );
 }

 /**
@@ -262,11 +444,10 @@ static int ib_srp_parse_lun ( const char *rp_comp,
 */
 static int ib_srp_parse_target_id_ext ( const char *rp_comp,
 					struct ib_srp_root_path *rp ) {
-	struct ib_srp_target_port_id *port_id =
-		ib_srp_target_port_id ( rp->port_ids );
+	union ib_srp_target_port_id *port_id = &rp->target;

-	return ib_srp_parse_byte_string ( rp_comp, port_id->id_ext.u.bytes,
-					  ( sizeof ( port_id->id_ext ) |
+	return ib_srp_parse_byte_string ( rp_comp, port_id->ib.id_ext.u.bytes,
+					  ( sizeof ( port_id->ib.id_ext ) |
 					    IB_SRP_PARSE_REQUIRED ) );
 }

@@ -279,11 +460,10 @@ static int ib_srp_parse_target_id_ext ( const char *rp_comp,
 */
 static int ib_srp_parse_target_ioc_guid ( const char *rp_comp,
 					  struct ib_srp_root_path *rp ) {
-	struct ib_srp_target_port_id *port_id =
-		ib_srp_target_port_id ( rp->port_ids );
+	union ib_srp_target_port_id *port_id = &rp->target;

-	return ib_srp_parse_byte_string ( rp_comp, port_id->ioc_guid.u.bytes,
-					  ( sizeof ( port_id->ioc_guid ) |
+	return ib_srp_parse_byte_string ( rp_comp, port_id->ib.ioc_guid.u.bytes,
+					  ( sizeof ( port_id->ib.ioc_guid ) |
 					    IB_SRP_PARSE_REQUIRED ) );
 }

@@ -301,7 +481,6 @@ struct ib_srp_root_path_parser {

 /** IB SRP root path components */
 static struct ib_srp_root_path_parser ib_srp_rp_parser[] = {
-	{ ib_srp_parse_literal },
 	{ ib_srp_parse_sgid },
 	{ ib_srp_parse_initiator_id_ext },
 	{ ib_srp_parse_initiator_hca_guid },
@@ -320,18 +499,13 @@ static struct ib_srp_root_path_parser ib_srp_rp_parser[] = {
 /**
 * Parse IB SRP root path
 *
- * @v srp		SRP device
- * @v rp_string		Root path
+ * @v rp_string		Root path string
+ * @v rp		IB SRP root path
 * @ret rc		Return status code
 */
-static int ib_srp_parse_root_path ( struct srp_device *srp,
-				    const char *rp_string ) {
-	struct ib_srp_parameters *ib_params = ib_srp_params ( srp );
-	struct ib_srp_root_path rp = {
-		.lun = &srp->lun,
-		.port_ids = &srp->port_ids,
-		.ib = ib_params,
-	};
+static int ib_srp_parse_root_path ( const char *rp_string,
+				    struct ib_srp_root_path *rp ) {
+	struct ib_srp_root_path_parser *parser;
 	char rp_string_copy[ strlen ( rp_string ) + 1 ];
 	char *rp_comp[IB_SRP_NUM_RP_COMPONENTS];
 	char *rp_string_tmp = rp_string_copy;
@@ -346,8 +520,8 @@ static int ib_srp_parse_root_path ( struct srp_device *srp,
 			break;
 		for ( ; *rp_string_tmp != ':' ; rp_string_tmp++ ) {
 			if ( ! *rp_string_tmp ) {
-				DBGC ( srp, "SRP %p root path \"%s\" too "
-				       "short\n", srp, rp_string );
+				DBG ( "IBSRP root path \"%s\" too short\n",
+				      rp_string );
 				return -EINVAL_RP_TOO_SHORT;
 			}
 		}
@@ -356,11 +530,11 @@ static int ib_srp_parse_root_path ( struct srp_device *srp,

 	/* Parse root path components */
 	for ( i = 0 ; i < IB_SRP_NUM_RP_COMPONENTS ; i++ ) {
-		if ( ( rc = ib_srp_rp_parser[i].parse ( rp_comp[i],
-							&rp ) ) != 0 ) {
-			DBGC ( srp, "SRP %p could not parse \"%s\" in root "
-			       "path \"%s\": %s\n", srp, rp_comp[i],
-			       rp_string, strerror ( rc ) );
+		parser = &ib_srp_rp_parser[i];
+		if ( ( rc = parser->parse ( rp_comp[i], rp ) ) != 0 ) {
+			DBG ( "IBSRP could not parse \"%s\" in root path "
+			      "\"%s\": %s\n", rp_comp[i], rp_string,
+			      strerror ( rc ) );
 			return rc;
 		}
 	}
@@ -369,41 +543,42 @@ static int ib_srp_parse_root_path ( struct srp_device *srp,
 }

 /**
- * Connect IB SRP session
+ * Open IB SRP URI
 *
- * @v srp		SRP device
+ * @v parent		Parent interface
+ * @v uri		URI
 * @ret rc		Return status code
 */
-static int ib_srp_connect ( struct srp_device *srp ) {
-	struct ib_srp_parameters *ib_params = ib_srp_params ( srp );
+static int ib_srp_open_uri ( struct interface *parent, struct uri *uri ) {
+	struct ib_srp_root_path rp;
 	struct ib_device *ibdev;
 	int rc;

+	/* Parse URI */
+	if ( ! uri->opaque )
+		return -EINVAL;
+	memset ( &rp, 0, sizeof ( rp ) );
+	if ( ( rc = ib_srp_parse_root_path ( uri->opaque, &rp ) ) != 0 )
+		return rc;
+
 	/* Identify Infiniband device */
-	ibdev = find_ibdev ( &ib_params->sgid );
+	ibdev = find_ibdev ( &rp.sgid );
 	if ( ! ibdev ) {
-		DBGC ( srp, "SRP %p could not identify Infiniband device\n",
-		       srp );
+		DBG ( "IBSRP could not identify Infiniband device\n" );
 		return -ENODEV;
 	}

-	/* Configure remaining SRP parameters */
-	srp->memory_handle = ibdev->rdma_key;
-
-	/* Open CMRC socket */
-	if ( ( rc = ib_cmrc_open ( &srp->socket, ibdev, &ib_params->dgid,
-				   &ib_params->service_id ) ) != 0 ) {
-		DBGC ( srp, "SRP %p could not open CMRC socket: %s\n",
-		       srp, strerror ( rc ) );
+	/* Open IB SRP device */
+	if ( ( rc = ib_srp_open ( parent, ibdev, &rp.dgid, &rp.service_id,
+				  &rp.initiator.srp, &rp.target.srp,
+				  &rp.lun ) ) != 0 )
 		return rc;
-	}

 	return 0;
 }

-/** IB SRP transport type */
-struct srp_transport_type ib_srp_transport = {
-	.priv_len = sizeof ( struct ib_srp_parameters ),
-	.parse_root_path = ib_srp_parse_root_path,
-	.connect = ib_srp_connect,
+/** IB SRP URI opener */
+struct uri_opener ib_srp_uri_opener __uri_opener = {
+	.scheme = "ib_srp",
+	.open = ib_srp_open_uri,
 };
--- a/src/net/tcp/iscsi.c
+++ b/src/net/tcp/iscsi.c
@@ -28,6 +28,7 @@ FILE_LICENCE ( GPL2_OR_LATER );
 #include <ipxe/vsprintf.h>
 #include <ipxe/socket.h>
 #include <ipxe/iobuf.h>
+#include <ipxe/uri.h>
 #include <ipxe/xfer.h>
 #include <ipxe/open.h>
 #include <ipxe/scsi.h>
@@ -38,6 +39,7 @@ FILE_LICENCE ( GPL2_OR_LATER );
 #include <ipxe/features.h>
 #include <ipxe/base16.h>
 #include <ipxe/base64.h>
+#include <ipxe/ibft.h>
 #include <ipxe/iscsi.h>

 /** @file
@@ -127,6 +129,36 @@ static void iscsi_rx_buffered_data_done ( struct iscsi_session *iscsi ) {
 	iscsi->rx_buffer = NULL;
 }

+/**
+ * Receive PDU data into buffer
+ *
+ * @v iscsi		iSCSI session
+ * @v data		Data to receive
+ * @v len		Length of data
+ * @ret rc		Return status code
+ *
+ * This can be used when the RX PDU type handler wishes to buffer up
+ * all received data and process the PDU as a single unit.  The caller
+ * is repsonsible for calling iscsi_rx_buffered_data_done() after
+ * processing the data.
+ */
+static int iscsi_rx_buffered_data ( struct iscsi_session *iscsi,
+				    const void *data, size_t len ) {
+
+	/* Allocate buffer on first call */
+	if ( ! iscsi->rx_buffer ) {
+		iscsi->rx_buffer = malloc ( iscsi->rx_len );
+		if ( ! iscsi->rx_buffer )
+			return -ENOMEM;
+	}
+
+	/* Copy data to buffer */
+	assert ( ( iscsi->rx_offset + len ) <= iscsi->rx_len );
+	memcpy ( ( iscsi->rx_buffer + iscsi->rx_offset ), data, len );
+
+	return 0;
+}
+
 /**
 * Free iSCSI session
 *
@@ -144,9 +176,44 @@ static void iscsi_free ( struct refcnt *refcnt ) {
 	free ( iscsi->target_password );
 	chap_finish ( &iscsi->chap );
 	iscsi_rx_buffered_data_done ( iscsi );
+	free ( iscsi->command );
 	free ( iscsi );
 }

+/**
+ * Shut down iSCSI interface
+ *
+ * @v iscsi		iSCSI session
+ * @v rc		Reason for close
+ */
+static void iscsi_close ( struct iscsi_session *iscsi, int rc ) {
+
+	/* A TCP graceful close is still an error from our point of view */
+	if ( rc == 0 )
+		rc = -ECONNRESET;
+
+	DBGC ( iscsi, "iSCSI %p closed: %s\n", iscsi, strerror ( rc ) );
+
+	/* Stop transmission process */
+	process_del ( &iscsi->process );
+
+	/* Shut down interfaces */
+	intf_shutdown ( &iscsi->socket, rc );
+	intf_shutdown ( &iscsi->control, rc );
+	intf_shutdown ( &iscsi->data, rc );
+}
+
+/**
+ * Assign new iSCSI initiator task tag
+ *
+ * @v iscsi		iSCSI session
+ */
+static void iscsi_new_itt ( struct iscsi_session *iscsi ) {
+	static uint16_t itt_idx;
+
+	iscsi->itt = ( ISCSI_TAG_MAGIC | (++itt_idx) );
+}
+
 /**
 * Open iSCSI transport-layer connection
 *
@@ -180,7 +247,7 @@ static int iscsi_open_connection ( struct iscsi_session *iscsi ) {
 		iscsi->status |= ISCSI_STATUS_AUTH_REVERSE_REQUIRED;

 	/* Assign fresh initiator task tag */
-	iscsi->itt++;
+	iscsi_new_itt ( iscsi );

 	/* Initiate login */
 	iscsi_start_login ( iscsi );
@@ -220,21 +287,34 @@ static void iscsi_close_connection ( struct iscsi_session *iscsi, int rc ) {
 *
 * @v iscsi		iSCSI session
 * @v rc		Return status code
+ * @v rsp		SCSI response, if any
 *
 * Note that iscsi_scsi_done() will not close the connection, and must
 * therefore be called only when the internal state machines are in an
 * appropriate state, otherwise bad things may happen on the next call
- * to iscsi_issue().  The general rule is to call iscsi_scsi_done()
- * only at the end of receiving a PDU; at this point the TX and RX
- * engines should both be idle.
+ * to iscsi_scsi_command().  The general rule is to call
+ * iscsi_scsi_done() only at the end of receiving a PDU; at this point
+ * the TX and RX engines should both be idle.
 */
-static void iscsi_scsi_done ( struct iscsi_session *iscsi, int rc ) {
+static void iscsi_scsi_done ( struct iscsi_session *iscsi, int rc,
+			      struct scsi_rsp *rsp ) {
+	uint32_t itt = iscsi->itt;

 	assert ( iscsi->tx_state == ISCSI_TX_IDLE );
-	assert ( iscsi->command != NULL );

-	iscsi->command->rc = rc;
+	/* Clear command */
+	free ( iscsi->command );
 	iscsi->command = NULL;
+
+	/* Send SCSI response, if any */
+	scsi_response ( &iscsi->data, rsp );
+
+	/* Close SCSI command, if this is still the same command.  (It
+	 * is possible that the command interface has already been
+	 * closed as a result of the SCSI response we sent.)
+	 */
+	if ( iscsi->itt == itt )
+		intf_restart ( &iscsi->data, rc );
 }

 /****************************************************************************
@@ -268,8 +348,9 @@ static void iscsi_start_command ( struct iscsi_session *iscsi ) {
 	if ( iscsi->command->data_out )
 		command->flags |= ISCSI_COMMAND_FLAG_WRITE;
 	/* lengths left as zero */
-	command->lun = iscsi->lun;
-	command->itt = htonl ( ++iscsi->itt );
+	memcpy ( &command->lun, &iscsi->command->lun,
+		 sizeof ( command->lun ) );
+	command->itt = htonl ( iscsi->itt );
 	command->exp_len = htonl ( iscsi->command->data_in_len |
 				   iscsi->command->data_out_len );
 	command->cmdsn = htonl ( iscsi->cmdsn );
@@ -297,28 +378,39 @@ static int iscsi_rx_scsi_response ( struct iscsi_session *iscsi,
 				    size_t remaining ) {
 	struct iscsi_bhs_scsi_response *response
 		= &iscsi->rx_bhs.scsi_response;
-	int sense_offset;
+	struct scsi_rsp rsp;
+	uint32_t residual_count;
+	int rc;

-	/* Capture the sense response code as it floats past, if present */
-	sense_offset = ISCSI_SENSE_RESPONSE_CODE_OFFSET - iscsi->rx_offset;
-	if ( ( sense_offset >= 0 ) && len ) {
-		iscsi->command->sense_response =
-			* ( ( char * ) data + sense_offset );
+	/* Buffer up the PDU data */
+	if ( ( rc = iscsi_rx_buffered_data ( iscsi, data, len ) ) != 0 ) {
+		DBGC ( iscsi, "iSCSI %p could not buffer login response: %s\n",
+		       iscsi, strerror ( rc ) );
+		return rc;
 	}
-
-	/* Wait for whole SCSI response to arrive */
 	if ( remaining )
 		return 0;
-	
-	/* Record SCSI status code */
-	iscsi->command->status = response->status;
+
+	/* Parse SCSI response and discard buffer */
+	memset ( &rsp, 0, sizeof ( rsp ) );
+	rsp.status = response->status;
+	residual_count = ntohl ( response->residual_count );
+	if ( response->flags & ISCSI_DATA_FLAG_OVERFLOW ) {
+		rsp.overrun = residual_count;
+	} else if ( response->flags & ISCSI_DATA_FLAG_UNDERFLOW ) {
+		rsp.overrun = -(residual_count);
+	}
+	if ( ISCSI_DATA_LEN ( response->lengths ) )
+		memcpy ( &rsp.sense, ( iscsi->rx_buffer + 2 ),
+			 sizeof ( rsp.sense ) );
+	iscsi_rx_buffered_data_done ( iscsi );

 	/* Check for errors */
 	if ( response->response != ISCSI_RESPONSE_COMMAND_COMPLETE )
 		return -EIO;

 	/* Mark as completed */
-	iscsi_scsi_done ( iscsi, 0 );
+	iscsi_scsi_done ( iscsi, 0, &rsp );
 	return 0;
 }

@@ -352,9 +444,8 @@ static int iscsi_rx_data_in ( struct iscsi_session *iscsi,
 	if ( data_in->flags & ISCSI_DATA_FLAG_STATUS ) {
 		assert ( ( offset + len ) == iscsi->command->data_in_len );
 		assert ( data_in->flags & ISCSI_FLAG_FINAL );
-		iscsi->command->status = data_in->status;
 		/* iSCSI cannot return an error status via a data-in */
-		iscsi_scsi_done ( iscsi, 0 );
+		iscsi_scsi_done ( iscsi, 0, NULL );
 	}

 	return 0;
@@ -412,7 +503,7 @@ static void iscsi_start_data_out ( struct iscsi_session *iscsi,
 	if ( len == remaining )
 		data_out->flags = ( ISCSI_FLAG_FINAL );
 	ISCSI_SET_LENGTHS ( data_out->lengths, 0, len );
-	data_out->lun = iscsi->lun;
+	data_out->lun = iscsi->command->lun;
 	data_out->itt = htonl ( iscsi->itt );
 	data_out->ttt = htonl ( iscsi->ttt );
 	data_out->expstatsn = htonl ( iscsi->statsn + 1 );
@@ -592,6 +683,19 @@ static void iscsi_start_login ( struct iscsi_session *iscsi ) {
 	struct iscsi_bhs_login_request *request = &iscsi->tx_bhs.login_request;
 	int len;

+	switch ( iscsi->status & ISCSI_LOGIN_CSG_MASK ) {
+	case ISCSI_LOGIN_CSG_SECURITY_NEGOTIATION:
+		DBGC ( iscsi, "iSCSI %p entering security negotiation\n",
+		       iscsi );
+		break;
+	case ISCSI_LOGIN_CSG_OPERATIONAL_NEGOTIATION:
+		DBGC ( iscsi, "iSCSI %p entering operational negotiation\n",
+		       iscsi );
+		break;
+	default:
+		assert ( 0 );
+	}
+
 	/* Construct BHS and initiate transmission */
 	iscsi_start_tx ( iscsi );
 	request->opcode = ( ISCSI_OPCODE_LOGIN_REQUEST |
@@ -604,7 +708,7 @@ static void iscsi_start_login ( struct iscsi_session *iscsi ) {
 	request->isid_iana_en = htonl ( ISCSI_ISID_IANA |
 					IANA_EN_FEN_SYSTEMS );
 	/* isid_iana_qual left as zero */
-	request->tsih = htons ( iscsi->tsih );
+	/* tsih left as zero */
 	request->itt = htonl ( iscsi->itt );
 	/* cid left as zero */
 	request->cmdsn = htonl ( iscsi->cmdsn );
@@ -1011,36 +1115,6 @@ static int iscsi_handle_strings ( struct iscsi_session *iscsi,
 	return 0;
 }

-/**
- * Receive PDU data into buffer
- *
- * @v iscsi		iSCSI session
- * @v data		Data to receive
- * @v len		Length of data
- * @ret rc		Return status code
- *
- * This can be used when the RX PDU type handler wishes to buffer up
- * all received data and process the PDU as a single unit.  The caller
- * is repsonsible for calling iscsi_rx_buffered_data_done() after
- * processing the data.
- */
-static int iscsi_rx_buffered_data ( struct iscsi_session *iscsi,
-				    const void *data, size_t len ) {
-
-	/* Allocate buffer on first call */
-	if ( ! iscsi->rx_buffer ) {
-		iscsi->rx_buffer = malloc ( iscsi->rx_len );
-		if ( ! iscsi->rx_buffer )
-			return -ENOMEM;
-	}
-
-	/* Copy data to buffer */
-	assert ( ( iscsi->rx_offset + len ) <= iscsi->rx_len );
-	memcpy ( ( iscsi->rx_buffer + iscsi->rx_offset ), data, len );
-
-	return 0;
-}
-
 /**
 * Convert iSCSI response status to return status code
 *
@@ -1119,7 +1193,6 @@ static int iscsi_rx_login_response ( struct iscsi_session *iscsi,
 		       response->status_class, response->status_detail );
 		rc = iscsi_status_to_rc ( response->status_class,
 					  response->status_detail );
-		iscsi->instant_rc = rc;
 		return rc;
 	}

@@ -1160,14 +1233,9 @@ static int iscsi_rx_login_response ( struct iscsi_session *iscsi,
 		return -EPROTO;
 	}

-	/* Reset retry count */
-	iscsi->retry_count = 0;
-
-	/* Record TSIH for future reference */
-	iscsi->tsih = ntohl ( response->tsih );
-	
-	/* Send the actual SCSI command */
-	iscsi_start_command ( iscsi );
+	/* Notify SCSI layer of window change */
+	DBGC ( iscsi, "iSCSI %p entering full feature phase\n", iscsi );
+	xfer_window_changed ( &iscsi->control );

 	return 0;
 }
@@ -1187,13 +1255,18 @@ static int iscsi_rx_login_response ( struct iscsi_session *iscsi,
 * be in transit at any one time.
 */
 static void iscsi_start_tx ( struct iscsi_session *iscsi ) {
+
 	assert ( iscsi->tx_state == ISCSI_TX_IDLE );
+	assert ( ! process_running ( &iscsi->process ) );
 	
 	/* Initialise TX BHS */
 	memset ( &iscsi->tx_bhs, 0, sizeof ( iscsi->tx_bhs ) );

 	/* Flag TX engine to start transmitting */
 	iscsi->tx_state = ISCSI_TX_BHS;
+
+	/* Start transmission process */
+	process_add ( &iscsi->process );
 }

 /**
@@ -1273,6 +1346,9 @@ static int iscsi_tx_data_padding ( struct iscsi_session *iscsi ) {
 static void iscsi_tx_done ( struct iscsi_session *iscsi ) {
 	struct iscsi_bhs_common *common = &iscsi->tx_bhs.common;

+	/* Stop transmission process */
+	process_del ( &iscsi->process );
+
 	switch ( common->opcode & ISCSI_OPCODE_MASK ) {
 	case ISCSI_OPCODE_DATA_OUT:
 		iscsi_data_out_done ( iscsi );
@@ -1305,9 +1381,6 @@ static void iscsi_tx_step ( struct process *process ) {
 	/* Select fragment to transmit */
 	while ( 1 ) {
 		switch ( iscsi->tx_state ) {
-		case ISCSI_TX_IDLE:
-			/* Stop processing */
-			return;
 		case ISCSI_TX_BHS:
 			tx = iscsi_tx_bhs;
 			tx_len = sizeof ( iscsi->tx_bhs );
@@ -1328,6 +1401,10 @@ static void iscsi_tx_step ( struct process *process ) {
 			tx_len = ISCSI_DATA_PAD_LEN ( common->lengths );
 			next_state = ISCSI_TX_IDLE;
 			break;
+		case ISCSI_TX_IDLE:
+			/* Stop processing */
+			iscsi_tx_done ( iscsi );
+			return;
 		default:
 			assert ( 0 );
 			return;
@@ -1343,13 +1420,13 @@ static void iscsi_tx_step ( struct process *process ) {
 		if ( ( rc = tx ( iscsi ) ) != 0 ) {
 			DBGC ( iscsi, "iSCSI %p could not transmit: %s\n",
 			       iscsi, strerror ( rc ) );
+			/* Transmission errors are fatal */
+			iscsi_close ( iscsi, rc );
 			return;
 		}

 		/* Move to next state */
 		iscsi->tx_state = next_state;
-		if ( next_state == ISCSI_TX_IDLE )
-			iscsi_tx_done ( iscsi );
 	}
 }

@@ -1495,8 +1572,6 @@ static int iscsi_socket_deliver ( struct iscsi_session *iscsi,
 				 remaining ) ) != 0 ) {
 			DBGC ( iscsi, "iSCSI %p could not process received "
 			       "data: %s\n", iscsi, strerror ( rc ) );
-			iscsi_close_connection ( iscsi, rc );
-			iscsi_scsi_done ( iscsi, rc );
 			goto done;
 		}

@@ -1518,41 +1593,14 @@ static int iscsi_socket_deliver ( struct iscsi_session *iscsi,
 done:
 	/* Free I/O buffer */
 	free_iob ( iobuf );
+
+	/* Destroy session on error */
+	if ( rc != 0 )
+		iscsi_close ( iscsi, rc );
+
 	return rc;
 }

-/**
- * Handle stream connection closure
- *
- * @v iscsi		iSCSI session
- * @v rc		Reason for close
- *
- */
-static void iscsi_socket_close ( struct iscsi_session *iscsi, int rc ) {
-
-	/* Even a graceful close counts as an error for iSCSI */
-	if ( ! rc )
-		rc = -ECONNRESET;
-
-	/* Close session cleanly */
-	iscsi_close_connection ( iscsi, rc );
-
-	/* Retry connection if within the retry limit, otherwise fail */
-	if ( ++iscsi->retry_count <= ISCSI_MAX_RETRIES ) {
-		DBGC ( iscsi, "iSCSI %p retrying connection (retry #%d)\n",
-		       iscsi, iscsi->retry_count );
-		if ( ( rc = iscsi_open_connection ( iscsi ) ) != 0 ) {
-			DBGC ( iscsi, "iSCSI %p could not reconnect: %s\n",
-			       iscsi, strerror ( rc ) );
-			iscsi_scsi_done ( iscsi, rc );
-		}
-	} else {
-		DBGC ( iscsi, "iSCSI %p retry count exceeded\n", iscsi );
-		iscsi->instant_rc = rc;
-		iscsi_scsi_done ( iscsi, rc );
-	}
-}
-
 /**
 * Handle redirection event
 *
@@ -1582,13 +1630,12 @@ static int iscsi_vredirect ( struct iscsi_session *iscsi, int type,

 	return xfer_vreopen ( &iscsi->socket, type, args );
 }
-			     

 /** iSCSI socket interface operations */
 static struct interface_operation iscsi_socket_operations[] = {
 	INTF_OP ( xfer_deliver, struct iscsi_session *, iscsi_socket_deliver ),
 	INTF_OP ( xfer_vredirect, struct iscsi_session *, iscsi_vredirect ),
-	INTF_OP ( intf_close, struct iscsi_session *, iscsi_socket_close ),
+	INTF_OP ( intf_close, struct iscsi_session *, iscsi_close ),
 };

 /** iSCSI socket interface descriptor */
@@ -1602,54 +1649,100 @@ static struct interface_descriptor iscsi_socket_desc =
 */

 /**
- * Issue SCSI command
+ * Check iSCSI flow-control window
 *
- * @v scsi		SCSI device
- * @v command		SCSI command
- * @ret rc		Return status code
+ * @v iscsi		iSCSI session
+ * @ret len		Length of window
 */
-static int iscsi_command ( struct scsi_device *scsi,
-			   struct scsi_command *command ) {
-	struct iscsi_session *iscsi =
-		container_of ( scsi->backend, struct iscsi_session, refcnt );
-	int rc;
+static size_t iscsi_scsi_window ( struct iscsi_session *iscsi ) {

-	/* Abort immediately if we have a recorded permanent failure */
-	if ( iscsi->instant_rc )
-		return iscsi->instant_rc;
-
-	/* Record SCSI command */
-	iscsi->command = command;
-
-	/* Issue command or open connection as appropriate */
-	if ( iscsi->status ) {
-		iscsi_start_command ( iscsi );
+	if ( ( ( iscsi->status & ISCSI_STATUS_PHASE_MASK ) ==
+	       ISCSI_STATUS_FULL_FEATURE_PHASE ) &&
+	     ( iscsi->command == NULL ) ) {
+		/* We cannot handle concurrent commands */
+		return 1;
 	} else {
-		if ( ( rc = iscsi_open_connection ( iscsi ) ) != 0 ) {
-			iscsi->command = NULL;
-			return rc;
-		}
+		return 0;
 	}
-
-	return 0;
 }

 /**
- * Shut down iSCSI interface
+ * Issue iSCSI SCSI command
 *
- * @v scsi		SCSI device
+ * @v iscsi		iSCSI session
+ * @v parent		Parent interface
+ * @v command		SCSI command
+ * @ret tag		Command tag, or negative error
 */
-void iscsi_detach ( struct scsi_device *scsi ) {
-	struct iscsi_session *iscsi =
-		container_of ( scsi->backend, struct iscsi_session, refcnt );
+static int iscsi_scsi_command ( struct iscsi_session *iscsi,
+				struct interface *parent,
+				struct scsi_cmd *command ) {

-	iscsi_close_connection ( iscsi, 0 );
-	process_del ( &iscsi->process );
-	scsi->command = scsi_detached_command;
-	ref_put ( scsi->backend );
-	scsi->backend = NULL;
+	/* This iSCSI implementation cannot handle multiple concurrent
+	 * commands or commands arriving before login is complete.
+	 */
+	if ( iscsi_scsi_window ( iscsi ) == 0 ) {
+		DBGC ( iscsi, "iSCSI %p cannot handle concurrent commands\n",
+		       iscsi );
+		return -EOPNOTSUPP;
+	}
+
+	/* Store command */
+	iscsi->command = malloc ( sizeof ( *command ) );
+	if ( ! iscsi->command )
+		return -ENOMEM;
+	memcpy ( iscsi->command, command, sizeof ( *command ) );
+
+	/* Assign new ITT */
+	iscsi_new_itt ( iscsi );
+
+	/* Start sending command */
+	iscsi_start_command ( iscsi );
+
+	/* Attach to parent interface and return */
+	intf_plug_plug ( &iscsi->data, parent );
+	return iscsi->itt;
 }

+/** iSCSI SCSI command-issuing interface operations */
+static struct interface_operation iscsi_control_op[] = {
+	INTF_OP ( scsi_command, struct iscsi_session *, iscsi_scsi_command ),
+	INTF_OP ( xfer_window, struct iscsi_session *, iscsi_scsi_window ),
+	INTF_OP ( intf_close, struct iscsi_session *, iscsi_close ),
+	INTF_OP ( acpi_describe, struct iscsi_session *, ibft_describe ),
+};
+
+/** iSCSI SCSI command-issuing interface descriptor */
+static struct interface_descriptor iscsi_control_desc =
+	INTF_DESC ( struct iscsi_session, control, iscsi_control_op );
+
+/**
+ * Close iSCSI command
+ *
+ * @v iscsi		iSCSI session
+ * @v rc		Reason for close
+ */
+static void iscsi_command_close ( struct iscsi_session *iscsi, int rc ) {
+
+	/* Restart interface */
+	intf_restart ( &iscsi->data, rc );
+
+	/* Treat unsolicited command closures mid-command as fatal,
+	 * because we have no code to handle partially-completed PDUs.
+	 */
+	if ( iscsi->command != NULL )
+		iscsi_close ( iscsi, ( ( rc == 0 ) ? -ECANCELED : rc ) );
+}
+
+/** iSCSI SCSI command interface operations */
+static struct interface_operation iscsi_data_op[] = {
+	INTF_OP ( intf_close, struct iscsi_session *, iscsi_command_close ),
+};
+
+/** iSCSI SCSI command interface descriptor */
+static struct interface_descriptor iscsi_data_desc =
+	INTF_DESC ( struct iscsi_session, data, iscsi_data_op );
+
 /****************************************************************************
 *
 * Instantiator
@@ -1658,8 +1751,7 @@ void iscsi_detach ( struct scsi_device *scsi ) {

 /** iSCSI root path components (as per RFC4173) */
 enum iscsi_root_path_component {
-	RP_LITERAL = 0,
-	RP_SERVERNAME,
+	RP_SERVERNAME = 0,
 	RP_PROTOCOL,
 	RP_PORT,
 	RP_LUN,
@@ -1779,60 +1871,95 @@ static int iscsi_set_auth ( struct iscsi_session *iscsi,
 }

 /**
- * Attach iSCSI interface
+ * Open iSCSI URI
 *
- * @v scsi		SCSI device
- * @v root_path		iSCSI root path (as per RFC4173)
+ * @v parent		Parent interface
+ * @v uri		URI
 * @ret rc		Return status code
 */
-int iscsi_attach ( struct scsi_device *scsi, const char *root_path ) {
+static int iscsi_open ( struct interface *parent, struct uri *uri ) {
 	struct iscsi_session *iscsi;
 	int rc;

+	/* Sanity check */
+	if ( ! uri->opaque ) {
+		rc = -EINVAL;
+		goto err_sanity_uri;
+	}
+
 	/* Allocate and initialise structure */
 	iscsi = zalloc ( sizeof ( *iscsi ) );
-	if ( ! iscsi )
-		return -ENOMEM;
+	if ( ! iscsi ) {
+		rc = -ENOMEM;
+		goto err_zalloc;
+	}
 	ref_init ( &iscsi->refcnt, iscsi_free );
+	intf_init ( &iscsi->control, &iscsi_control_desc, &iscsi->refcnt );
+	intf_init ( &iscsi->data, &iscsi_data_desc, &iscsi->refcnt );
 	intf_init ( &iscsi->socket, &iscsi_socket_desc, &iscsi->refcnt );
-	process_init ( &iscsi->process, iscsi_tx_step, &iscsi->refcnt );
+	process_init_stopped ( &iscsi->process, iscsi_tx_step,
+			       &iscsi->refcnt );

 	/* Parse root path */
-	if ( ( rc = iscsi_parse_root_path ( iscsi, root_path ) ) != 0 )
-		goto err;
+	if ( ( rc = iscsi_parse_root_path ( iscsi, uri->opaque ) ) != 0 )
+		goto err_parse_root_path;
 	/* Set fields not specified by root path */
 	if ( ( rc = iscsi_set_auth ( iscsi,
 				     iscsi_initiator_username,
 				     iscsi_initiator_password,
 				     iscsi_target_username,
 				     iscsi_target_password ) ) != 0 )
-		goto err;
+		goto err_set_auth;

 	/* Sanity checks */
 	if ( ! iscsi->target_address ) {
 		DBGC ( iscsi, "iSCSI %p does not yet support discovery\n",
 		       iscsi );
 		rc = -ENOTSUP_DISCOVERY;
-		goto err;
+		goto err_sanity_address;
 	}
 	if ( ! iscsi->target_iqn ) {
 		DBGC ( iscsi, "iSCSI %p no target address supplied in %s\n",
-		       iscsi, root_path );
+		       iscsi, uri->opaque );
 		rc = -EINVAL;
-		goto err;
+		goto err_sanity_iqn;
 	}

-	/* Attach parent interface, mortalise self, and return */
-	scsi->backend = ref_get ( &iscsi->refcnt );
-	scsi->command = iscsi_command;
+	/* Open socket */
+	if ( ( rc = iscsi_open_connection ( iscsi ) ) != 0 )
+		goto err_open_connection;
+
+	/* Attach SCSI device to parent interface */
+	if ( ( rc = scsi_open ( parent, &iscsi->control,
+				&iscsi->lun ) ) != 0 ) {
+		DBGC ( iscsi, "iSCSI %p could not create SCSI device: %s\n",
+		       iscsi, strerror ( rc ) );
+		goto err_scsi_open;
+	}
+
+	/* Mortalise self, and return */
 	ref_put ( &iscsi->refcnt );
 	return 0;
 	
- err:
+ err_scsi_open:
+ err_open_connection:
+ err_sanity_iqn:
+ err_sanity_address:
+ err_set_auth:
+ err_parse_root_path:
+	iscsi_close ( iscsi, rc );
 	ref_put ( &iscsi->refcnt );
+ err_zalloc:
+ err_sanity_uri:
 	return rc;
 }

+/** iSCSI URI opener */
+struct uri_opener iscsi_uri_opener __uri_opener = {
+	.scheme = "iscsi",
+	.open = iscsi_open,
+};
+
 /****************************************************************************
 *
 * Settings