Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
tjh.dev
kernel
os
linux
1.. SPDX-License-Identifier: GPL-2.0
2
3=================
4SCSI FC Transport
5=================
6
7Date: 11/18/2008
8
9Kernel Revisions for features::
10
11 rports : <<TBS>>
12 vports : 2.6.22
13 bsg support : 2.6.30 (?TBD?)
14
15
16Introduction
17============
18This file documents the features and components of the SCSI FC Transport.
19It also provides documents the API between the transport and FC LLDDs.
20
21The FC transport can be found at::
22
23 drivers/scsi/scsi_transport_fc.c
24 include/scsi/scsi_transport_fc.h
25 include/scsi/scsi_netlink_fc.h
26 include/scsi/scsi_bsg_fc.h
27
28This file is found at Documentation/scsi/scsi_fc_transport.rst
29
30
31FC Remote Ports (rports)
32========================
33
34 In the Fibre Channel (FC) subsystem, a remote port (rport) refers to a
35 remote Fibre Channel node that the local port can communicate with.
36 These are typically storage targets (e.g., arrays, tapes) that respond
37 to SCSI commands over FC transport.
38
39 In Linux, rports are managed by the FC transport class and are
40 represented in sysfs under:
41
42 /sys/class/fc_remote_ports/
43
44 Each rport directory contains attributes describing the remote port,
45 such as port ID, node name, port state, and link speed.
46
47 rports are typically created by the FC transport when a new device is
48 discovered during a fabric login or scan, and they persist until the
49 device is removed or the link is lost.
50
51 Common attributes:
52 - node_name: World Wide Node Name (WWNN).
53 - port_name: World Wide Port Name (WWPN).
54 - port_id: FC address of the remote port.
55 - roles: Indicates if the port is an initiator, target, or both.
56 - port_state: Shows the current operational state.
57
58 After discovering a remote port, the driver typically populates a
59 fc_rport_identifiers structure and invokes fc_remote_port_add() to
60 create and register the remote port with the SCSI subsystem via the
61 Fibre Channel (FC) transport class.
62
63 rports are also visible via sysfs as children of the FC host adapter.
64
65 For developers: use fc_remote_port_add() and fc_remote_port_delete() when
66 implementing a driver that interacts with the FC transport class.
67
68
69FC Virtual Ports (vports)
70=========================
71
72Overview
73--------
74
75 New FC standards have defined mechanisms which allows for a single physical
76 port to appear on as multiple communication ports. Using the N_Port Id
77 Virtualization (NPIV) mechanism, a point-to-point connection to a Fabric
78 can be assigned more than 1 N_Port_ID. Each N_Port_ID appears as a
79 separate port to other endpoints on the fabric, even though it shares one
80 physical link to the switch for communication. Each N_Port_ID can have a
81 unique view of the fabric based on fabric zoning and array lun-masking
82 (just like a normal non-NPIV adapter). Using the Virtual Fabric (VF)
83 mechanism, adding a fabric header to each frame allows the port to
84 interact with the Fabric Port to join multiple fabrics. The port will
85 obtain an N_Port_ID on each fabric it joins. Each fabric will have its
86 own unique view of endpoints and configuration parameters. NPIV may be
87 used together with VF so that the port can obtain multiple N_Port_IDs
88 on each virtual fabric.
89
90 The FC transport is now recognizing a new object - a vport. A vport is
91 an entity that has a world-wide unique World Wide Port Name (wwpn) and
92 World Wide Node Name (wwnn). The transport also allows for the FC4's to
93 be specified for the vport, with FCP_Initiator being the primary role
94 expected. Once instantiated by one of the above methods, it will have a
95 distinct N_Port_ID and view of fabric endpoints and storage entities.
96 The fc_host associated with the physical adapter will export the ability
97 to create vports. The transport will create the vport object within the
98 Linux device tree, and instruct the fc_host's driver to instantiate the
99 virtual port. Typically, the driver will create a new scsi_host instance
100 on the vport, resulting in a unique <H,C,T,L> namespace for the vport.
101 Thus, whether a FC port is based on a physical port or on a virtual port,
102 each will appear as a unique scsi_host with its own target and lun space.
103
104 .. Note::
105 At this time, the transport is written to create only NPIV-based
106 vports. However, consideration was given to VF-based vports and it
107 should be a minor change to add support if needed. The remaining
108 discussion will concentrate on NPIV.
109
110 .. Note::
111 World Wide Name assignment (and uniqueness guarantees) are left
112 up to an administrative entity controlling the vport. For example,
113 if vports are to be associated with virtual machines, a XEN mgmt
114 utility would be responsible for creating wwpn/wwnn's for the vport,
115 using its own naming authority and OUI. (Note: it already does this
116 for virtual MAC addresses).
117
118
119Device Trees and Vport Objects:
120-------------------------------
121
122 Today, the device tree typically contains the scsi_host object,
123 with rports and scsi target objects underneath it. Currently the FC
124 transport creates the vport object and places it under the scsi_host
125 object corresponding to the physical adapter. The LLDD will allocate
126 a new scsi_host for the vport and link its object under the vport.
127 The remainder of the tree under the vports scsi_host is the same
128 as the non-NPIV case. The transport is written currently to easily
129 allow the parent of the vport to be something other than the scsi_host.
130 This could be used in the future to link the object onto a vm-specific
131 device tree. If the vport's parent is not the physical port's scsi_host,
132 a symbolic link to the vport object will be placed in the physical
133 port's scsi_host.
134
135 Here's what to expect in the device tree :
136
137 The typical Physical Port's Scsi_Host::
138
139 /sys/devices/.../host17/
140
141 and it has the typical descendant tree::
142
143 /sys/devices/.../host17/rport-17:0-0/target17:0:0/17:0:0:0:
144
145 and then the vport is created on the Physical Port::
146
147 /sys/devices/.../host17/vport-17:0-0
148
149 and the vport's Scsi_Host is then created::
150
151 /sys/devices/.../host17/vport-17:0-0/host18
152
153 and then the rest of the tree progresses, such as::
154
155 /sys/devices/.../host17/vport-17:0-0/host18/rport-18:0-0/target18:0:0/18:0:0:0:
156
157 Here's what to expect in the sysfs tree::
158
159 scsi_hosts:
160 /sys/class/scsi_host/host17 physical port's scsi_host
161 /sys/class/scsi_host/host18 vport's scsi_host
162 fc_hosts:
163 /sys/class/fc_host/host17 physical port's fc_host
164 /sys/class/fc_host/host18 vport's fc_host
165 fc_vports:
166 /sys/class/fc_vports/vport-17:0-0 the vport's fc_vport
167 fc_rports:
168 /sys/class/fc_remote_ports/rport-17:0-0 rport on the physical port
169 /sys/class/fc_remote_ports/rport-18:0-0 rport on the vport
170
171
172Vport Attributes
173----------------
174
175 The new fc_vport class object has the following attributes
176
177 node_name: Read_Only
178 The WWNN of the vport
179
180 port_name: Read_Only
181 The WWPN of the vport
182
183 roles: Read_Only
184 Indicates the FC4 roles enabled on the vport.
185
186 symbolic_name: Read_Write
187 A string, appended to the driver's symbolic port name string, which
188 is registered with the switch to identify the vport. For example,
189 a hypervisor could set this string to "Xen Domain 2 VM 5 Vport 2",
190 and this set of identifiers can be seen on switch management screens
191 to identify the port.
192
193 vport_delete: Write_Only
194 When written with a "1", will tear down the vport.
195
196 vport_disable: Write_Only
197 When written with a "1", will transition the vport to a disabled.
198 state. The vport will still be instantiated with the Linux kernel,
199 but it will not be active on the FC link.
200 When written with a "0", will enable the vport.
201
202 vport_last_state: Read_Only
203 Indicates the previous state of the vport. See the section below on
204 "Vport States".
205
206 vport_state: Read_Only
207 Indicates the state of the vport. See the section below on
208 "Vport States".
209
210 vport_type: Read_Only
211 Reflects the FC mechanism used to create the virtual port.
212 Only NPIV is supported currently.
213
214
215 For the fc_host class object, the following attributes are added for vports:
216
217 max_npiv_vports: Read_Only
218 Indicates the maximum number of NPIV-based vports that the
219 driver/adapter can support on the fc_host.
220
221 npiv_vports_inuse: Read_Only
222 Indicates how many NPIV-based vports have been instantiated on the
223 fc_host.
224
225 vport_create: Write_Only
226 A "simple" create interface to instantiate a vport on an fc_host.
227 A "<WWPN>:<WWNN>" string is written to the attribute. The transport
228 then instantiates the vport object and calls the LLDD to create the
229 vport with the role of FCP_Initiator. Each WWN is specified as 16
230 hex characters and may *not* contain any prefixes (e.g. 0x, x, etc).
231
232 vport_delete: Write_Only
233 A "simple" delete interface to teardown a vport. A "<WWPN>:<WWNN>"
234 string is written to the attribute. The transport will locate the
235 vport on the fc_host with the same WWNs and tear it down. Each WWN
236 is specified as 16 hex characters and may *not* contain any prefixes
237 (e.g. 0x, x, etc).
238
239
240Vport States
241------------
242
243 Vport instantiation consists of two parts:
244
245 - Creation with the kernel and LLDD. This means all transport and
246 driver data structures are built up, and device objects created.
247 This is equivalent to a driver "attach" on an adapter, which is
248 independent of the adapter's link state.
249 - Instantiation of the vport on the FC link via ELS traffic, etc.
250 This is equivalent to a "link up" and successful link initialization.
251
252 Further information can be found in the interfaces section below for
253 Vport Creation.
254
255 Once a vport has been instantiated with the kernel/LLDD, a vport state
256 can be reported via the sysfs attribute. The following states exist:
257
258 FC_VPORT_UNKNOWN - Unknown
259 An temporary state, typically set only while the vport is being
260 instantiated with the kernel and LLDD.
261
262 FC_VPORT_ACTIVE - Active
263 The vport has been successfully been created on the FC link.
264 It is fully functional.
265
266 FC_VPORT_DISABLED - Disabled
267 The vport instantiated, but "disabled". The vport is not instantiated
268 on the FC link. This is equivalent to a physical port with the
269 link "down".
270
271 FC_VPORT_LINKDOWN - Linkdown
272 The vport is not operational as the physical link is not operational.
273
274 FC_VPORT_INITIALIZING - Initializing
275 The vport is in the process of instantiating on the FC link.
276 The LLDD will set this state just prior to starting the ELS traffic
277 to create the vport. This state will persist until the vport is
278 successfully created (state becomes FC_VPORT_ACTIVE) or it fails
279 (state is one of the values below). As this state is transitory,
280 it will not be preserved in the "vport_last_state".
281
282 FC_VPORT_NO_FABRIC_SUPP - No Fabric Support
283 The vport is not operational. One of the following conditions were
284 encountered:
285
286 - The FC topology is not Point-to-Point
287 - The FC port is not connected to an F_Port
288 - The F_Port has indicated that NPIV is not supported.
289
290 FC_VPORT_NO_FABRIC_RSCS - No Fabric Resources
291 The vport is not operational. The Fabric failed FDISC with a status
292 indicating that it does not have sufficient resources to complete
293 the operation.
294
295 FC_VPORT_FABRIC_LOGOUT - Fabric Logout
296 The vport is not operational. The Fabric has LOGO'd the N_Port_ID
297 associated with the vport.
298
299 FC_VPORT_FABRIC_REJ_WWN - Fabric Rejected WWN
300 The vport is not operational. The Fabric failed FDISC with a status
301 indicating that the WWN's are not valid.
302
303 FC_VPORT_FAILED - VPort Failed
304 The vport is not operational. This is a catchall for all other
305 error conditions.
306
307
308 The following state table indicates the different state transitions:
309
310 +------------------+--------------------------------+---------------------+
311 | State | Event | New State |
312 +==================+================================+=====================+
313 | n/a | Initialization | Unknown |
314 +------------------+--------------------------------+---------------------+
315 | Unknown: | Link Down | Linkdown |
316 | +--------------------------------+---------------------+
317 | | Link Up & Loop | No Fabric Support |
318 | +--------------------------------+---------------------+
319 | | Link Up & no Fabric | No Fabric Support |
320 | +--------------------------------+---------------------+
321 | | Link Up & FLOGI response | No Fabric Support |
322 | | indicates no NPIV support | |
323 | +--------------------------------+---------------------+
324 | | Link Up & FDISC being sent | Initializing |
325 | +--------------------------------+---------------------+
326 | | Disable request | Disable |
327 +------------------+--------------------------------+---------------------+
328 | Linkdown: | Link Up | Unknown |
329 +------------------+--------------------------------+---------------------+
330 | Initializing: | FDISC ACC | Active |
331 | +--------------------------------+---------------------+
332 | | FDISC LS_RJT w/ no resources | No Fabric Resources |
333 | +--------------------------------+---------------------+
334 | | FDISC LS_RJT w/ invalid | Fabric Rejected WWN |
335 | | pname or invalid nport_id | |
336 | +--------------------------------+---------------------+
337 | | FDISC LS_RJT failed for | Vport Failed |
338 | | other reasons | |
339 | +--------------------------------+---------------------+
340 | | Link Down | Linkdown |
341 | +--------------------------------+---------------------+
342 | | Disable request | Disable |
343 +------------------+--------------------------------+---------------------+
344 | Disable: | Enable request | Unknown |
345 +------------------+--------------------------------+---------------------+
346 | Active: | LOGO received from fabric | Fabric Logout |
347 | +--------------------------------+---------------------+
348 | | Link Down | Linkdown |
349 | +--------------------------------+---------------------+
350 | | Disable request | Disable |
351 +------------------+--------------------------------+---------------------+
352 | Fabric Logout: | Link still up | Unknown |
353 +------------------+--------------------------------+---------------------+
354
355The following 4 error states all have the same transitions::
356
357 No Fabric Support:
358 No Fabric Resources:
359 Fabric Rejected WWN:
360 Vport Failed:
361 Disable request Disable
362 Link goes down Linkdown
363
364
365Transport <-> LLDD Interfaces
366-----------------------------
367
368Vport support by LLDD:
369
370 The LLDD indicates support for vports by supplying a vport_create()
371 function in the transport template. The presence of this function will
372 cause the creation of the new attributes on the fc_host. As part of
373 the physical port completing its initialization relative to the
374 transport, it should set the max_npiv_vports attribute to indicate the
375 maximum number of vports the driver and/or adapter supports.
376
377
378Vport Creation:
379
380 The LLDD vport_create() syntax is::
381
382 int vport_create(struct fc_vport *vport, bool disable)
383
384 where:
385
386 ======= ===========================================================
387 vport Is the newly allocated vport object
388 disable If "true", the vport is to be created in a disabled stated.
389 If "false", the vport is to be enabled upon creation.
390 ======= ===========================================================
391
392 When a request is made to create a new vport (via sgio/netlink, or the
393 vport_create fc_host attribute), the transport will validate that the LLDD
394 can support another vport (e.g. max_npiv_vports > npiv_vports_inuse).
395 If not, the create request will be failed. If space remains, the transport
396 will increment the vport count, create the vport object, and then call the
397 LLDD's vport_create() function with the newly allocated vport object.
398
399 As mentioned above, vport creation is divided into two parts:
400
401 - Creation with the kernel and LLDD. This means all transport and
402 driver data structures are built up, and device objects created.
403 This is equivalent to a driver "attach" on an adapter, which is
404 independent of the adapter's link state.
405 - Instantiation of the vport on the FC link via ELS traffic, etc.
406 This is equivalent to a "link up" and successful link initialization.
407
408 The LLDD's vport_create() function will not synchronously wait for both
409 parts to be fully completed before returning. It must validate that the
410 infrastructure exists to support NPIV, and complete the first part of
411 vport creation (data structure build up) before returning. We do not
412 hinge vport_create() on the link-side operation mainly because:
413
414 - The link may be down. It is not a failure if it is. It simply
415 means the vport is in an inoperable state until the link comes up.
416 This is consistent with the link bouncing post vport creation.
417 - The vport may be created in a disabled state.
418 - This is consistent with a model where: the vport equates to a
419 FC adapter. The vport_create is synonymous with driver attachment
420 to the adapter, which is independent of link state.
421
422 .. Note::
423
424 special error codes have been defined to delineate infrastructure
425 failure cases for quicker resolution.
426
427 The expected behavior for the LLDD's vport_create() function is:
428
429 - Validate Infrastructure:
430
431 - If the driver or adapter cannot support another vport, whether
432 due to improper firmware, (a lie about) max_npiv, or a lack of
433 some other resource - return VPCERR_UNSUPPORTED.
434 - If the driver validates the WWN's against those already active on
435 the adapter and detects an overlap - return VPCERR_BAD_WWN.
436 - If the driver detects the topology is loop, non-fabric, or the
437 FLOGI did not support NPIV - return VPCERR_NO_FABRIC_SUPP.
438
439 - Allocate data structures. If errors are encountered, such as out
440 of memory conditions, return the respective negative Exxx error code.
441 - If the role is FCP Initiator, the LLDD is to :
442
443 - Call scsi_host_alloc() to allocate a scsi_host for the vport.
444 - Call scsi_add_host(new_shost, &vport->dev) to start the scsi_host
445 and bind it as a child of the vport device.
446 - Initializes the fc_host attribute values.
447
448 - Kick of further vport state transitions based on the disable flag and
449 link state - and return success (zero).
450
451 LLDD Implementers Notes:
452
453 - It is suggested that there be a different fc_function_templates for
454 the physical port and the virtual port. The physical port's template
455 would have the vport_create, vport_delete, and vport_disable functions,
456 while the vports would not.
457 - It is suggested that there be different scsi_host_templates
458 for the physical port and virtual port. Likely, there are driver
459 attributes, embedded into the scsi_host_template, that are applicable
460 for the physical port only (link speed, topology setting, etc). This
461 ensures that the attributes are applicable to the respective scsi_host.
462
463
464Vport Disable/Enable:
465
466 The LLDD vport_disable() syntax is::
467
468 int vport_disable(struct fc_vport *vport, bool disable)
469
470 where:
471
472 ======= =======================================
473 vport Is vport to be enabled or disabled
474 disable If "true", the vport is to be disabled.
475 If "false", the vport is to be enabled.
476 ======= =======================================
477
478 When a request is made to change the disabled state on a vport, the
479 transport will validate the request against the existing vport state.
480 If the request is to disable and the vport is already disabled, the
481 request will fail. Similarly, if the request is to enable, and the
482 vport is not in a disabled state, the request will fail. If the request
483 is valid for the vport state, the transport will call the LLDD to
484 change the vport's state.
485
486 Within the LLDD, if a vport is disabled, it remains instantiated with
487 the kernel and LLDD, but it is not active or visible on the FC link in
488 any way. (see Vport Creation and the 2 part instantiation discussion).
489 The vport will remain in this state until it is deleted or re-enabled.
490 When enabling a vport, the LLDD reinstantiates the vport on the FC
491 link - essentially restarting the LLDD statemachine (see Vport States
492 above).
493
494
495Vport Deletion:
496
497 The LLDD vport_delete() syntax is::
498
499 int vport_delete(struct fc_vport *vport)
500
501 where:
502
503 vport: Is vport to delete
504
505 When a request is made to delete a vport (via sgio/netlink, or via the
506 fc_host or fc_vport vport_delete attributes), the transport will call
507 the LLDD to terminate the vport on the FC link, and teardown all other
508 datastructures and references. If the LLDD completes successfully,
509 the transport will teardown the vport objects and complete the vport
510 removal. If the LLDD delete request fails, the vport object will remain,
511 but will be in an indeterminate state.
512
513 Within the LLDD, the normal code paths for a scsi_host teardown should
514 be followed. E.g. If the vport has a FCP Initiator role, the LLDD
515 will call fc_remove_host() for the vports scsi_host, followed by
516 scsi_remove_host() and scsi_host_put() for the vports scsi_host.
517
518
519Other:
520 fc_host port_type attribute:
521 There is a new fc_host port_type value - FC_PORTTYPE_NPIV. This value
522 must be set on all vport-based fc_hosts. Normally, on a physical port,
523 the port_type attribute would be set to NPORT, NLPORT, etc based on the
524 topology type and existence of the fabric. As this is not applicable to
525 a vport, it makes more sense to report the FC mechanism used to create
526 the vport.
527
528 Driver unload:
529 FC drivers are required to call fc_remove_host() prior to calling
530 scsi_remove_host(). This allows the fc_host to tear down all remote
531 ports prior the scsi_host being torn down. The fc_remove_host() call
532 was updated to remove all vports for the fc_host as well.
533
534
535Transport supplied functions
536----------------------------
537
538The following functions are supplied by the FC-transport for use by LLDs.
539
540 ================== =========================
541 fc_vport_create create a vport
542 fc_vport_terminate detach and remove a vport
543 ================== =========================
544
545Details::
546
547 /**
548 * fc_vport_create - Admin App or LLDD requests creation of a vport
549 * @shost: scsi host the virtual port is connected to.
550 * @ids: The world wide names, FC4 port roles, etc for
551 * the virtual port.
552 *
553 * Notes:
554 * This routine assumes no locks are held on entry.
555 */
556 struct fc_vport *
557 fc_vport_create(struct Scsi_Host *shost, struct fc_vport_identifiers *ids)
558
559 /**
560 * fc_vport_terminate - Admin App or LLDD requests termination of a vport
561 * @vport: fc_vport to be terminated
562 *
563 * Calls the LLDD vport_delete() function, then deallocates and removes
564 * the vport from the shost and object tree.
565 *
566 * Notes:
567 * This routine assumes no locks are held on entry.
568 */
569 int
570 fc_vport_terminate(struct fc_vport *vport)
571
572
573FC BSG support (CT & ELS passthru, and more)
574============================================
575
576<< To Be Supplied >>
577
578
579
580
581
582Credits
583=======
584The following people have contributed to this document:
585
586
587
588
589
590
591James Smart
592james.smart@broadcom.com
593