linux/include/scsi/scsi_transport_fc.h

531 lines
17 KiB
C
Raw Normal View History

/*
* FiberChannel transport specific attributes exported to sysfs.
*
* Copyright (c) 2003 Silicon Graphics, Inc. All rights reserved.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
* ========
*
* Copyright (C) 2004-2005 James Smart, Emulex Corporation
* Rewrite for host, target, device, and remote port attributes,
* statistics, and service functions...
*
*/
#ifndef SCSI_TRANSPORT_FC_H
#define SCSI_TRANSPORT_FC_H
#include <linux/sched.h>
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-18 16:03:35 +00:00
#include <scsi/scsi.h>
struct scsi_transport_template;
/*
* FC Port definitions - Following FC HBAAPI guidelines
*
* Note: Not all binary values for the different fields match HBAAPI.
* Instead, we use densely packed ordinal values or enums.
* We get away with this as we never present the actual binary values
* externally. For sysfs, we always present the string that describes
* the value. Thus, an admin doesn't need a magic HBAAPI decoder ring
* to understand the values. The HBAAPI user-space library is free to
* convert the strings into the HBAAPI-specified binary values.
*
* Note: Not all HBAAPI-defined values are contained in the definitions
* below. Those not appropriate to an fc_host (e.g. FCP initiator) have
* been removed.
*/
/*
* fc_port_type: If you alter this, you also need to alter scsi_transport_fc.c
* (for the ascii descriptions).
*/
enum fc_port_type {
FC_PORTTYPE_UNKNOWN,
FC_PORTTYPE_OTHER,
FC_PORTTYPE_NOTPRESENT,
FC_PORTTYPE_NPORT, /* Attached to FPort */
FC_PORTTYPE_NLPORT, /* (Public) Loop w/ FLPort */
FC_PORTTYPE_LPORT, /* (Private) Loop w/o FLPort */
FC_PORTTYPE_PTP, /* Point to Point w/ another NPort */
};
/*
* fc_port_state: If you alter this, you also need to alter scsi_transport_fc.c
* (for the ascii descriptions).
*/
enum fc_port_state {
FC_PORTSTATE_UNKNOWN,
FC_PORTSTATE_NOTPRESENT,
FC_PORTSTATE_ONLINE,
FC_PORTSTATE_OFFLINE, /* User has taken Port Offline */
FC_PORTSTATE_BLOCKED,
FC_PORTSTATE_BYPASSED,
FC_PORTSTATE_DIAGNOSTICS,
FC_PORTSTATE_LINKDOWN,
FC_PORTSTATE_ERROR,
FC_PORTSTATE_LOOPBACK,
FC_PORTSTATE_DELETED,
};
/*
* FC Classes of Service
* Note: values are not enumerated, as they can be "or'd" together
* for reporting (e.g. report supported_classes). If you alter this list,
* you also need to alter scsi_transport_fc.c (for the ascii descriptions).
*/
#define FC_COS_UNSPECIFIED 0
#define FC_COS_CLASS1 2
#define FC_COS_CLASS2 4
#define FC_COS_CLASS3 8
#define FC_COS_CLASS4 0x10
#define FC_COS_CLASS6 0x40
/*
* FC Port Speeds
* Note: values are not enumerated, as they can be "or'd" together
* for reporting (e.g. report supported_speeds). If you alter this list,
* you also need to alter scsi_transport_fc.c (for the ascii descriptions).
*/
#define FC_PORTSPEED_UNKNOWN 0 /* Unknown - transceiver
incapable of reporting */
#define FC_PORTSPEED_1GBIT 1
#define FC_PORTSPEED_2GBIT 2
#define FC_PORTSPEED_4GBIT 4
#define FC_PORTSPEED_10GBIT 8
#define FC_PORTSPEED_NOT_NEGOTIATED (1 << 15) /* Speed not established */
/*
* fc_tgtid_binding_type: If you alter this, you also need to alter
* scsi_transport_fc.c (for the ascii descriptions).
*/
enum fc_tgtid_binding_type {
FC_TGTID_BIND_NONE,
FC_TGTID_BIND_BY_WWPN,
FC_TGTID_BIND_BY_WWNN,
FC_TGTID_BIND_BY_ID,
};
/*
* FC Remote Port Roles
* Note: values are not enumerated, as they can be "or'd" together
* for reporting (e.g. report roles). If you alter this list,
* you also need to alter scsi_transport_fc.c (for the ascii descriptions).
*/
#define FC_RPORT_ROLE_UNKNOWN 0x00
#define FC_RPORT_ROLE_FCP_TARGET 0x01
#define FC_RPORT_ROLE_FCP_INITIATOR 0x02
#define FC_RPORT_ROLE_IP_PORT 0x04
/*
* fc_rport_identifiers: This set of data contains all elements
* to uniquely identify a remote FC port. The driver uses this data
* to report the existence of a remote FC port in the topology. Internally,
* the transport uses this data for attributes and to manage consistent
* target id bindings.
*/
struct fc_rport_identifiers {
u64 node_name;
u64 port_name;
u32 port_id;
u32 roles;
};
/* Macro for use in defining Remote Port attributes */
#define FC_RPORT_ATTR(_name,_mode,_show,_store) \
struct class_device_attribute class_device_attr_rport_##_name = \
__ATTR(_name,_mode,_show,_store)
/*
* FC Remote Port Attributes
*
* This structure exists for each remote FC port that a LLDD notifies
* the subsystem of. A remote FC port may or may not be a SCSI Target,
* also be a SCSI initiator, IP endpoint, etc. As such, the remote
* port is considered a separate entity, independent of "role" (such
* as scsi target).
*
* --
*
* Attributes are based on HBAAPI V2.0 definitions. Only those
* attributes that are determinable by the local port (aka Host)
* are contained.
*
* Fixed attributes are not expected to change. The driver is
* expected to set these values after successfully calling
* fc_remote_port_add(). The transport fully manages all get functions
* w/o driver interaction.
*
* Dynamic attributes are expected to change. The driver participates
* in all get/set operations via functions provided by the driver.
*
* Private attributes are transport-managed values. They are fully
* managed by the transport w/o driver interaction.
*/
struct fc_rport { /* aka fc_starget_attrs */
/* Fixed Attributes */
u32 maxframe_size;
u32 supported_classes;
/* Dynamic Attributes */
u32 dev_loss_tmo; /* Remote Port loss timeout in seconds. */
/* Private (Transport-managed) Attributes */
u64 node_name;
u64 port_name;
u32 port_id;
u32 roles;
enum fc_port_state port_state; /* Will only be ONLINE or UNKNOWN */
u32 scsi_target_id;
/* exported data */
void *dd_data; /* Used for driver-specific storage */
/* internal data */
unsigned int channel;
u32 number;
[SCSI] FC transport: fixes for workq deadlocks As previously reported via Michael Reed, the FC transport took a hit in 2.6.15 (perhaps a little earlier) when we solved a recursion error. There are 2 deadlocks occurring: - With scan and the delete items sharing the same workq, flushing the workq for the delete code was getting it stalled behind a very long running scan code path. - There's a deadlock where scsi_remove_target() has to sit behind scsi_scan_target() due to contention over the scan_lock(). This patch resolves the 1st deadlock and significantly reduces the odds of the second. So far, we have only replicated the 2nd deadlock on a highly-parallel SMP system. More on the 2nd deadlock in a following email. This patch reworks the transport to: - Only use the scsi host workq for scanning - Use 2 other workq's internally. One for deletions, the other for scheduled deletions. Originally, we tried this with a single workq, but the occassional flushes of the scheduled queues was hitting the second deadlock with a slightly higher frequency. In the future, we'll look at the LLDD's and the transport to see if we can get rid of this extra overhead. - When moving to the other workq's we tightened up some object states and some lock handling. - Properly syncs adds/deletes - minor code cleanups - directly reference fc_host_attrs, rather than through attribute macros - flush the right workq on delayed work cancel failures. Large kudos to Michael Reed who has been working this issue for the last month. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-04-10 14:14:05 +00:00
u8 flags;
struct list_head peers;
struct device dev;
struct work_struct dev_loss_work;
struct work_struct scan_work;
[SCSI] FC transport: fixes for workq deadlocks As previously reported via Michael Reed, the FC transport took a hit in 2.6.15 (perhaps a little earlier) when we solved a recursion error. There are 2 deadlocks occurring: - With scan and the delete items sharing the same workq, flushing the workq for the delete code was getting it stalled behind a very long running scan code path. - There's a deadlock where scsi_remove_target() has to sit behind scsi_scan_target() due to contention over the scan_lock(). This patch resolves the 1st deadlock and significantly reduces the odds of the second. So far, we have only replicated the 2nd deadlock on a highly-parallel SMP system. More on the 2nd deadlock in a following email. This patch reworks the transport to: - Only use the scsi host workq for scanning - Use 2 other workq's internally. One for deletions, the other for scheduled deletions. Originally, we tried this with a single workq, but the occassional flushes of the scheduled queues was hitting the second deadlock with a slightly higher frequency. In the future, we'll look at the LLDD's and the transport to see if we can get rid of this extra overhead. - When moving to the other workq's we tightened up some object states and some lock handling. - Properly syncs adds/deletes - minor code cleanups - directly reference fc_host_attrs, rather than through attribute macros - flush the right workq on delayed work cancel failures. Large kudos to Michael Reed who has been working this issue for the last month. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-04-10 14:14:05 +00:00
struct work_struct stgt_delete_work;
struct work_struct rport_delete_work;
} __attribute__((aligned(sizeof(unsigned long))));
[SCSI] FC transport: fixes for workq deadlocks As previously reported via Michael Reed, the FC transport took a hit in 2.6.15 (perhaps a little earlier) when we solved a recursion error. There are 2 deadlocks occurring: - With scan and the delete items sharing the same workq, flushing the workq for the delete code was getting it stalled behind a very long running scan code path. - There's a deadlock where scsi_remove_target() has to sit behind scsi_scan_target() due to contention over the scan_lock(). This patch resolves the 1st deadlock and significantly reduces the odds of the second. So far, we have only replicated the 2nd deadlock on a highly-parallel SMP system. More on the 2nd deadlock in a following email. This patch reworks the transport to: - Only use the scsi host workq for scanning - Use 2 other workq's internally. One for deletions, the other for scheduled deletions. Originally, we tried this with a single workq, but the occassional flushes of the scheduled queues was hitting the second deadlock with a slightly higher frequency. In the future, we'll look at the LLDD's and the transport to see if we can get rid of this extra overhead. - When moving to the other workq's we tightened up some object states and some lock handling. - Properly syncs adds/deletes - minor code cleanups - directly reference fc_host_attrs, rather than through attribute macros - flush the right workq on delayed work cancel failures. Large kudos to Michael Reed who has been working this issue for the last month. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-04-10 14:14:05 +00:00
/* bit field values for struct fc_rport "flags" field: */
#define FC_RPORT_DEVLOSS_PENDING 0x01
#define FC_RPORT_SCAN_PENDING 0x02
#define dev_to_rport(d) \
container_of(d, struct fc_rport, dev)
#define transport_class_to_rport(classdev) \
dev_to_rport(classdev->dev)
#define rport_to_shost(r) \
dev_to_shost(r->dev.parent)
/*
* FC SCSI Target Attributes
*
* The SCSI Target is considered an extention of a remote port (as
* a remote port can be more than a SCSI Target). Within the scsi
* subsystem, we leave the Target as a separate entity. Doing so
* provides backward compatibility with prior FC transport api's,
* and lets remote ports be handled entirely within the FC transport
* and independently from the scsi subsystem. The drawback is that
* some data will be duplicated.
*/
struct fc_starget_attrs { /* aka fc_target_attrs */
/* Dynamic Attributes */
u64 node_name;
u64 port_name;
u32 port_id;
};
#define fc_starget_node_name(x) \
(((struct fc_starget_attrs *)&(x)->starget_data)->node_name)
#define fc_starget_port_name(x) \
(((struct fc_starget_attrs *)&(x)->starget_data)->port_name)
#define fc_starget_port_id(x) \
(((struct fc_starget_attrs *)&(x)->starget_data)->port_id)
#define starget_to_rport(s) \
scsi_is_fc_rport(s->dev.parent) ? dev_to_rport(s->dev.parent) : NULL
/*
* FC Local Port (Host) Statistics
*/
/* FC Statistics - Following FC HBAAPI v2.0 guidelines */
struct fc_host_statistics {
/* port statistics */
u64 seconds_since_last_reset;
u64 tx_frames;
u64 tx_words;
u64 rx_frames;
u64 rx_words;
u64 lip_count;
u64 nos_count;
u64 error_frames;
u64 dumped_frames;
u64 link_failure_count;
u64 loss_of_sync_count;
u64 loss_of_signal_count;
u64 prim_seq_protocol_err_count;
u64 invalid_tx_word_count;
u64 invalid_crc_count;
/* fc4 statistics (only FCP supported currently) */
u64 fcp_input_requests;
u64 fcp_output_requests;
u64 fcp_control_requests;
u64 fcp_input_megabytes;
u64 fcp_output_megabytes;
};
/*
* FC Local Port (Host) Attributes
*
* Attributes are based on HBAAPI V2.0 definitions.
* Note: OSDeviceName is determined by user-space library
*
* Fixed attributes are not expected to change. The driver is
* expected to set these values after successfully calling scsi_add_host().
* The transport fully manages all get functions w/o driver interaction.
*
* Dynamic attributes are expected to change. The driver participates
* in all get/set operations via functions provided by the driver.
*
* Private attributes are transport-managed values. They are fully
* managed by the transport w/o driver interaction.
*/
#define FC_FC4_LIST_SIZE 32
#define FC_SYMBOLIC_NAME_SIZE 256
#define FC_VERSION_STRING_SIZE 64
#define FC_SERIAL_NUMBER_SIZE 80
struct fc_host_attrs {
/* Fixed Attributes */
u64 node_name;
u64 port_name;
u64 permanent_port_name;
u32 supported_classes;
u8 supported_fc4s[FC_FC4_LIST_SIZE];
u32 supported_speeds;
u32 maxframe_size;
char serial_number[FC_SERIAL_NUMBER_SIZE];
/* Dynamic Attributes */
u32 port_id;
enum fc_port_type port_type;
enum fc_port_state port_state;
u8 active_fc4s[FC_FC4_LIST_SIZE];
u32 speed;
u64 fabric_name;
char symbolic_name[FC_SYMBOLIC_NAME_SIZE];
char system_hostname[FC_SYMBOLIC_NAME_SIZE];
/* Private (Transport-managed) Attributes */
enum fc_tgtid_binding_type tgtid_bind_type;
/* internal data */
struct list_head rports;
struct list_head rport_bindings;
u32 next_rport_number;
u32 next_target_id;
[SCSI] FC transport: fixes for workq deadlocks As previously reported via Michael Reed, the FC transport took a hit in 2.6.15 (perhaps a little earlier) when we solved a recursion error. There are 2 deadlocks occurring: - With scan and the delete items sharing the same workq, flushing the workq for the delete code was getting it stalled behind a very long running scan code path. - There's a deadlock where scsi_remove_target() has to sit behind scsi_scan_target() due to contention over the scan_lock(). This patch resolves the 1st deadlock and significantly reduces the odds of the second. So far, we have only replicated the 2nd deadlock on a highly-parallel SMP system. More on the 2nd deadlock in a following email. This patch reworks the transport to: - Only use the scsi host workq for scanning - Use 2 other workq's internally. One for deletions, the other for scheduled deletions. Originally, we tried this with a single workq, but the occassional flushes of the scheduled queues was hitting the second deadlock with a slightly higher frequency. In the future, we'll look at the LLDD's and the transport to see if we can get rid of this extra overhead. - When moving to the other workq's we tightened up some object states and some lock handling. - Properly syncs adds/deletes - minor code cleanups - directly reference fc_host_attrs, rather than through attribute macros - flush the right workq on delayed work cancel failures. Large kudos to Michael Reed who has been working this issue for the last month. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-04-10 14:14:05 +00:00
/* work queues for rport state manipulation */
char work_q_name[KOBJ_NAME_LEN];
struct workqueue_struct *work_q;
char devloss_work_q_name[KOBJ_NAME_LEN];
struct workqueue_struct *devloss_work_q;
};
[SCSI] FC transport: fixes for workq deadlocks As previously reported via Michael Reed, the FC transport took a hit in 2.6.15 (perhaps a little earlier) when we solved a recursion error. There are 2 deadlocks occurring: - With scan and the delete items sharing the same workq, flushing the workq for the delete code was getting it stalled behind a very long running scan code path. - There's a deadlock where scsi_remove_target() has to sit behind scsi_scan_target() due to contention over the scan_lock(). This patch resolves the 1st deadlock and significantly reduces the odds of the second. So far, we have only replicated the 2nd deadlock on a highly-parallel SMP system. More on the 2nd deadlock in a following email. This patch reworks the transport to: - Only use the scsi host workq for scanning - Use 2 other workq's internally. One for deletions, the other for scheduled deletions. Originally, we tried this with a single workq, but the occassional flushes of the scheduled queues was hitting the second deadlock with a slightly higher frequency. In the future, we'll look at the LLDD's and the transport to see if we can get rid of this extra overhead. - When moving to the other workq's we tightened up some object states and some lock handling. - Properly syncs adds/deletes - minor code cleanups - directly reference fc_host_attrs, rather than through attribute macros - flush the right workq on delayed work cancel failures. Large kudos to Michael Reed who has been working this issue for the last month. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-04-10 14:14:05 +00:00
#define shost_to_fc_host(x) \
((struct fc_host_attrs *)(x)->shost_data)
#define fc_host_node_name(x) \
(((struct fc_host_attrs *)(x)->shost_data)->node_name)
#define fc_host_port_name(x) \
(((struct fc_host_attrs *)(x)->shost_data)->port_name)
#define fc_host_permanent_port_name(x) \
(((struct fc_host_attrs *)(x)->shost_data)->permanent_port_name)
#define fc_host_supported_classes(x) \
(((struct fc_host_attrs *)(x)->shost_data)->supported_classes)
#define fc_host_supported_fc4s(x) \
(((struct fc_host_attrs *)(x)->shost_data)->supported_fc4s)
#define fc_host_supported_speeds(x) \
(((struct fc_host_attrs *)(x)->shost_data)->supported_speeds)
#define fc_host_maxframe_size(x) \
(((struct fc_host_attrs *)(x)->shost_data)->maxframe_size)
#define fc_host_serial_number(x) \
(((struct fc_host_attrs *)(x)->shost_data)->serial_number)
#define fc_host_port_id(x) \
(((struct fc_host_attrs *)(x)->shost_data)->port_id)
#define fc_host_port_type(x) \
(((struct fc_host_attrs *)(x)->shost_data)->port_type)
#define fc_host_port_state(x) \
(((struct fc_host_attrs *)(x)->shost_data)->port_state)
#define fc_host_active_fc4s(x) \
(((struct fc_host_attrs *)(x)->shost_data)->active_fc4s)
#define fc_host_speed(x) \
(((struct fc_host_attrs *)(x)->shost_data)->speed)
#define fc_host_fabric_name(x) \
(((struct fc_host_attrs *)(x)->shost_data)->fabric_name)
#define fc_host_symbolic_name(x) \
(((struct fc_host_attrs *)(x)->shost_data)->symbolic_name)
#define fc_host_system_hostname(x) \
(((struct fc_host_attrs *)(x)->shost_data)->system_hostname)
#define fc_host_tgtid_bind_type(x) \
(((struct fc_host_attrs *)(x)->shost_data)->tgtid_bind_type)
#define fc_host_rports(x) \
(((struct fc_host_attrs *)(x)->shost_data)->rports)
#define fc_host_rport_bindings(x) \
(((struct fc_host_attrs *)(x)->shost_data)->rport_bindings)
#define fc_host_next_rport_number(x) \
(((struct fc_host_attrs *)(x)->shost_data)->next_rport_number)
#define fc_host_next_target_id(x) \
(((struct fc_host_attrs *)(x)->shost_data)->next_target_id)
[SCSI] FC transport: fixes for workq deadlocks As previously reported via Michael Reed, the FC transport took a hit in 2.6.15 (perhaps a little earlier) when we solved a recursion error. There are 2 deadlocks occurring: - With scan and the delete items sharing the same workq, flushing the workq for the delete code was getting it stalled behind a very long running scan code path. - There's a deadlock where scsi_remove_target() has to sit behind scsi_scan_target() due to contention over the scan_lock(). This patch resolves the 1st deadlock and significantly reduces the odds of the second. So far, we have only replicated the 2nd deadlock on a highly-parallel SMP system. More on the 2nd deadlock in a following email. This patch reworks the transport to: - Only use the scsi host workq for scanning - Use 2 other workq's internally. One for deletions, the other for scheduled deletions. Originally, we tried this with a single workq, but the occassional flushes of the scheduled queues was hitting the second deadlock with a slightly higher frequency. In the future, we'll look at the LLDD's and the transport to see if we can get rid of this extra overhead. - When moving to the other workq's we tightened up some object states and some lock handling. - Properly syncs adds/deletes - minor code cleanups - directly reference fc_host_attrs, rather than through attribute macros - flush the right workq on delayed work cancel failures. Large kudos to Michael Reed who has been working this issue for the last month. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-04-10 14:14:05 +00:00
#define fc_host_work_q_name(x) \
(((struct fc_host_attrs *)(x)->shost_data)->work_q_name)
#define fc_host_work_q(x) \
(((struct fc_host_attrs *)(x)->shost_data)->work_q)
#define fc_host_devloss_work_q_name(x) \
(((struct fc_host_attrs *)(x)->shost_data)->devloss_work_q_name)
#define fc_host_devloss_work_q(x) \
(((struct fc_host_attrs *)(x)->shost_data)->devloss_work_q)
/* The functions by which the transport class and the driver communicate */
struct fc_function_template {
void (*get_rport_dev_loss_tmo)(struct fc_rport *);
void (*set_rport_dev_loss_tmo)(struct fc_rport *, u32);
void (*get_starget_node_name)(struct scsi_target *);
void (*get_starget_port_name)(struct scsi_target *);
void (*get_starget_port_id)(struct scsi_target *);
void (*get_host_port_id)(struct Scsi_Host *);
void (*get_host_port_type)(struct Scsi_Host *);
void (*get_host_port_state)(struct Scsi_Host *);
void (*get_host_active_fc4s)(struct Scsi_Host *);
void (*get_host_speed)(struct Scsi_Host *);
void (*get_host_fabric_name)(struct Scsi_Host *);
void (*get_host_symbolic_name)(struct Scsi_Host *);
void (*set_host_system_hostname)(struct Scsi_Host *);
struct fc_host_statistics * (*get_fc_host_stats)(struct Scsi_Host *);
void (*reset_fc_host_stats)(struct Scsi_Host *);
int (*issue_fc_host_lip)(struct Scsi_Host *);
/* allocation lengths for host-specific data */
u32 dd_fcrport_size;
/*
* The driver sets these to tell the transport class it
* wants the attributes displayed in sysfs. If the show_ flag
* is not set, the attribute will be private to the transport
* class
*/
/* remote port fixed attributes */
unsigned long show_rport_maxframe_size:1;
unsigned long show_rport_supported_classes:1;
unsigned long show_rport_dev_loss_tmo:1;
/*
* target dynamic attributes
* These should all be "1" if the driver uses the remote port
* add/delete functions (so attributes reflect rport values).
*/
unsigned long show_starget_node_name:1;
unsigned long show_starget_port_name:1;
unsigned long show_starget_port_id:1;
/* host fixed attributes */
unsigned long show_host_node_name:1;
unsigned long show_host_port_name:1;
unsigned long show_host_permanent_port_name:1;
unsigned long show_host_supported_classes:1;
unsigned long show_host_supported_fc4s:1;
unsigned long show_host_supported_speeds:1;
unsigned long show_host_maxframe_size:1;
unsigned long show_host_serial_number:1;
/* host dynamic attributes */
unsigned long show_host_port_id:1;
unsigned long show_host_port_type:1;
unsigned long show_host_port_state:1;
unsigned long show_host_active_fc4s:1;
unsigned long show_host_speed:1;
unsigned long show_host_fabric_name:1;
unsigned long show_host_symbolic_name:1;
unsigned long show_host_system_hostname:1;
};
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-18 16:03:35 +00:00
/**
* fc_remote_port_chkready - called to validate the remote port state
* prior to initiating io to the port.
*
* Returns a scsi result code that can be returned by the LLDD.
*
* @rport: remote port to be checked
**/
static inline int
fc_remote_port_chkready(struct fc_rport *rport)
{
int result;
switch (rport->port_state) {
case FC_PORTSTATE_ONLINE:
[SCSI] FC transport: fixes for workq deadlocks As previously reported via Michael Reed, the FC transport took a hit in 2.6.15 (perhaps a little earlier) when we solved a recursion error. There are 2 deadlocks occurring: - With scan and the delete items sharing the same workq, flushing the workq for the delete code was getting it stalled behind a very long running scan code path. - There's a deadlock where scsi_remove_target() has to sit behind scsi_scan_target() due to contention over the scan_lock(). This patch resolves the 1st deadlock and significantly reduces the odds of the second. So far, we have only replicated the 2nd deadlock on a highly-parallel SMP system. More on the 2nd deadlock in a following email. This patch reworks the transport to: - Only use the scsi host workq for scanning - Use 2 other workq's internally. One for deletions, the other for scheduled deletions. Originally, we tried this with a single workq, but the occassional flushes of the scheduled queues was hitting the second deadlock with a slightly higher frequency. In the future, we'll look at the LLDD's and the transport to see if we can get rid of this extra overhead. - When moving to the other workq's we tightened up some object states and some lock handling. - Properly syncs adds/deletes - minor code cleanups - directly reference fc_host_attrs, rather than through attribute macros - flush the right workq on delayed work cancel failures. Large kudos to Michael Reed who has been working this issue for the last month. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-04-10 14:14:05 +00:00
if (rport->roles & FC_RPORT_ROLE_FCP_TARGET)
result = 0;
else if (rport->flags & FC_RPORT_DEVLOSS_PENDING)
result = DID_IMM_RETRY << 16;
else
result = DID_NO_CONNECT << 16;
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-18 16:03:35 +00:00
break;
case FC_PORTSTATE_BLOCKED:
[SCSI] FC transport: fixes for workq deadlocks As previously reported via Michael Reed, the FC transport took a hit in 2.6.15 (perhaps a little earlier) when we solved a recursion error. There are 2 deadlocks occurring: - With scan and the delete items sharing the same workq, flushing the workq for the delete code was getting it stalled behind a very long running scan code path. - There's a deadlock where scsi_remove_target() has to sit behind scsi_scan_target() due to contention over the scan_lock(). This patch resolves the 1st deadlock and significantly reduces the odds of the second. So far, we have only replicated the 2nd deadlock on a highly-parallel SMP system. More on the 2nd deadlock in a following email. This patch reworks the transport to: - Only use the scsi host workq for scanning - Use 2 other workq's internally. One for deletions, the other for scheduled deletions. Originally, we tried this with a single workq, but the occassional flushes of the scheduled queues was hitting the second deadlock with a slightly higher frequency. In the future, we'll look at the LLDD's and the transport to see if we can get rid of this extra overhead. - When moving to the other workq's we tightened up some object states and some lock handling. - Properly syncs adds/deletes - minor code cleanups - directly reference fc_host_attrs, rather than through attribute macros - flush the right workq on delayed work cancel failures. Large kudos to Michael Reed who has been working this issue for the last month. Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2006-04-10 14:14:05 +00:00
result = DID_IMM_RETRY << 16;
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-18 16:03:35 +00:00
break;
default:
result = DID_NO_CONNECT << 16;
break;
}
return result;
}
static inline u64 wwn_to_u64(u8 *wwn)
{
return (u64)wwn[0] << 56 | (u64)wwn[1] << 48 |
(u64)wwn[2] << 40 | (u64)wwn[3] << 32 |
(u64)wwn[4] << 24 | (u64)wwn[5] << 16 |
(u64)wwn[6] << 8 | (u64)wwn[7];
}
static inline void u64_to_wwn(u64 inm, u8 *wwn)
{
wwn[0] = (inm >> 56) & 0xff;
wwn[1] = (inm >> 48) & 0xff;
wwn[2] = (inm >> 40) & 0xff;
wwn[3] = (inm >> 32) & 0xff;
wwn[4] = (inm >> 24) & 0xff;
wwn[5] = (inm >> 16) & 0xff;
wwn[6] = (inm >> 8) & 0xff;
wwn[7] = inm & 0xff;
}
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-18 16:03:35 +00:00
struct scsi_transport_template *fc_attach_transport(
struct fc_function_template *);
void fc_release_transport(struct scsi_transport_template *);
void fc_remove_host(struct Scsi_Host *);
struct fc_rport *fc_remote_port_add(struct Scsi_Host *shost,
int channel, struct fc_rport_identifiers *ids);
void fc_remote_port_delete(struct fc_rport *rport);
void fc_remote_port_rolechg(struct fc_rport *rport, u32 roles);
int scsi_is_fc_rport(const struct device *);
#endif /* SCSI_TRANSPORT_FC_H */