Root/Documentation/scsi/scsi_fc_transport.txt

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

Archive Download this file



interactive