linux/drivers/scsi/fcoe/fcoe_ctlr.c

3284 lines
88 KiB
C
Raw Normal View History

/*
* Copyright (c) 2008-2009 Cisco Systems, Inc. All rights reserved.
* Copyright (c) 2009 Intel Corporation. All rights reserved.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope 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.,
* 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
*
* Maintained at www.Open-FCoE.org
*/
#include <linux/types.h>
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/timer.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/ethtool.h>
#include <linux/if_ether.h>
#include <linux/if_vlan.h>
#include <linux/errno.h>
#include <linux/bitops.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 16:04:11 +08:00
#include <linux/slab.h>
#include <net/rtnetlink.h>
#include <scsi/fc/fc_els.h>
#include <scsi/fc/fc_fs.h>
#include <scsi/fc/fc_fip.h>
#include <scsi/fc/fc_encaps.h>
#include <scsi/fc/fc_fcoe.h>
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
#include <scsi/fc/fc_fcp.h>
#include <scsi/libfc.h>
#include <scsi/libfcoe.h>
#include "libfcoe.h"
#define FCOE_CTLR_MIN_FKA 500 /* min keep alive (mS) */
#define FCOE_CTLR_DEF_FKA FIP_DEF_FKA /* default keep alive (mS) */
treewide: setup_timer() -> timer_setup() This converts all remaining cases of the old setup_timer() API into using timer_setup(), where the callback argument is the structure already holding the struct timer_list. These should have no behavioral changes, since they just change which pointer is passed into the callback with the same available pointers after conversion. It handles the following examples, in addition to some other variations. Casting from unsigned long: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... setup_timer(&ptr->my_timer, my_callback, ptr); and forced object casts: void my_callback(struct something *ptr) { ... } ... setup_timer(&ptr->my_timer, my_callback, (unsigned long)ptr); become: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... timer_setup(&ptr->my_timer, my_callback, 0); Direct function assignments: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... ptr->my_timer.function = my_callback; have a temporary cast added, along with converting the args: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... ptr->my_timer.function = (TIMER_FUNC_TYPE)my_callback; And finally, callbacks without a data assignment: void my_callback(unsigned long data) { ... } ... setup_timer(&ptr->my_timer, my_callback, 0); have their argument renamed to verify they're unused during conversion: void my_callback(struct timer_list *unused) { ... } ... timer_setup(&ptr->my_timer, my_callback, 0); The conversion is done with the following Coccinelle script: spatch --very-quiet --all-includes --include-headers \ -I ./arch/x86/include -I ./arch/x86/include/generated \ -I ./include -I ./arch/x86/include/uapi \ -I ./arch/x86/include/generated/uapi -I ./include/uapi \ -I ./include/generated/uapi --include ./include/linux/kconfig.h \ --dir . \ --cocci-file ~/src/data/timer_setup.cocci @fix_address_of@ expression e; @@ setup_timer( -&(e) +&e , ...) // Update any raw setup_timer() usages that have a NULL callback, but // would otherwise match change_timer_function_usage, since the latter // will update all function assignments done in the face of a NULL // function initialization in setup_timer(). @change_timer_function_usage_NULL@ expression _E; identifier _timer; type _cast_data; @@ ( -setup_timer(&_E->_timer, NULL, _E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E->_timer, NULL, (_cast_data)_E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E._timer, NULL, &_E); +timer_setup(&_E._timer, NULL, 0); | -setup_timer(&_E._timer, NULL, (_cast_data)&_E); +timer_setup(&_E._timer, NULL, 0); ) @change_timer_function_usage@ expression _E; identifier _timer; struct timer_list _stl; identifier _callback; type _cast_func, _cast_data; @@ ( -setup_timer(&_E->_timer, _callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | _E->_timer@_stl.function = _callback; | _E->_timer@_stl.function = &_callback; | _E->_timer@_stl.function = (_cast_func)_callback; | _E->_timer@_stl.function = (_cast_func)&_callback; | _E._timer@_stl.function = _callback; | _E._timer@_stl.function = &_callback; | _E._timer@_stl.function = (_cast_func)_callback; | _E._timer@_stl.function = (_cast_func)&_callback; ) // callback(unsigned long arg) @change_callback_handle_cast depends on change_timer_function_usage@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; identifier _handle; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { ( ... when != _origarg _handletype *_handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg ) } // callback(unsigned long arg) without existing variable @change_callback_handle_cast_no_arg depends on change_timer_function_usage && !change_callback_handle_cast@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { + _handletype *_origarg = from_timer(_origarg, t, _timer); + ... when != _origarg - (_handletype *)_origarg + _origarg ... when != _origarg } // Avoid already converted callbacks. @match_callback_converted depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier t; @@ void _callback(struct timer_list *t) { ... } // callback(struct something *handle) @change_callback_handle_arg depends on change_timer_function_usage && !match_callback_converted && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; @@ void _callback( -_handletype *_handle +struct timer_list *t ) { + _handletype *_handle = from_timer(_handle, t, _timer); ... } // If change_callback_handle_arg ran on an empty function, remove // the added handler. @unchange_callback_handle_arg depends on change_timer_function_usage && change_callback_handle_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; identifier t; @@ void _callback(struct timer_list *t) { - _handletype *_handle = from_timer(_handle, t, _timer); } // We only want to refactor the setup_timer() data argument if we've found // the matching callback. This undoes changes in change_timer_function_usage. @unchange_timer_function_usage depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg && !change_callback_handle_arg@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type change_timer_function_usage._cast_data; @@ ( -timer_setup(&_E->_timer, _callback, 0); +setup_timer(&_E->_timer, _callback, (_cast_data)_E); | -timer_setup(&_E._timer, _callback, 0); +setup_timer(&_E._timer, _callback, (_cast_data)&_E); ) // If we fixed a callback from a .function assignment, fix the // assignment cast now. @change_timer_function_assignment depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_func; typedef TIMER_FUNC_TYPE; @@ ( _E->_timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -&_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)_callback; +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -&_callback; +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; ) // Sometimes timer functions are called directly. Replace matched args. @change_timer_function_calls depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression _E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_data; @@ _callback( ( -(_cast_data)_E +&_E->_timer | -(_cast_data)&_E +&_E._timer | -_E +&_E->_timer ) ) // If a timer has been configured without a data argument, it can be // converted without regard to the callback argument, since it is unused. @match_timer_function_unused_data@ expression _E; identifier _timer; identifier _callback; @@ ( -setup_timer(&_E->_timer, _callback, 0); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0L); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0UL); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0L); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0UL); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_timer, _callback, 0); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0L); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0UL); +timer_setup(&_timer, _callback, 0); | -setup_timer(_timer, _callback, 0); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0L); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0UL); +timer_setup(_timer, _callback, 0); ) @change_callback_unused_data depends on match_timer_function_unused_data@ identifier match_timer_function_unused_data._callback; type _origtype; identifier _origarg; @@ void _callback( -_origtype _origarg +struct timer_list *unused ) { ... when != _origarg } Signed-off-by: Kees Cook <keescook@chromium.org>
2017-10-17 05:43:17 +08:00
static void fcoe_ctlr_timeout(struct timer_list *);
static void fcoe_ctlr_timer_work(struct work_struct *);
static void fcoe_ctlr_recv_work(struct work_struct *);
static int fcoe_ctlr_flogi_retry(struct fcoe_ctlr *);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
static void fcoe_ctlr_vn_start(struct fcoe_ctlr *);
static int fcoe_ctlr_vn_recv(struct fcoe_ctlr *, struct sk_buff *);
static void fcoe_ctlr_vn_timeout(struct fcoe_ctlr *);
static int fcoe_ctlr_vn_lookup(struct fcoe_ctlr *, u32, u8 *);
static int fcoe_ctlr_vlan_recv(struct fcoe_ctlr *, struct sk_buff *);
static u8 fcoe_all_fcfs[ETH_ALEN] = FIP_ALL_FCF_MACS;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
static u8 fcoe_all_enode[ETH_ALEN] = FIP_ALL_ENODE_MACS;
static u8 fcoe_all_vn2vn[ETH_ALEN] = FIP_ALL_VN2VN_MACS;
static u8 fcoe_all_p2p[ETH_ALEN] = FIP_ALL_P2P_MACS;
static const char * const fcoe_ctlr_states[] = {
[FIP_ST_DISABLED] = "DISABLED",
[FIP_ST_LINK_WAIT] = "LINK_WAIT",
[FIP_ST_AUTO] = "AUTO",
[FIP_ST_NON_FIP] = "NON_FIP",
[FIP_ST_ENABLED] = "ENABLED",
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
[FIP_ST_VNMP_START] = "VNMP_START",
[FIP_ST_VNMP_PROBE1] = "VNMP_PROBE1",
[FIP_ST_VNMP_PROBE2] = "VNMP_PROBE2",
[FIP_ST_VNMP_CLAIM] = "VNMP_CLAIM",
[FIP_ST_VNMP_UP] = "VNMP_UP",
};
static const char *fcoe_ctlr_state(enum fip_state state)
{
const char *cp = "unknown";
if (state < ARRAY_SIZE(fcoe_ctlr_states))
cp = fcoe_ctlr_states[state];
if (!cp)
cp = "unknown";
return cp;
}
/**
* fcoe_ctlr_set_state() - Set and do debug printing for the new FIP state.
* @fip: The FCoE controller
* @state: The new state
*/
static void fcoe_ctlr_set_state(struct fcoe_ctlr *fip, enum fip_state state)
{
if (state == fip->state)
return;
if (fip->lp)
LIBFCOE_FIP_DBG(fip, "state %s -> %s\n",
fcoe_ctlr_state(fip->state), fcoe_ctlr_state(state));
fip->state = state;
}
/**
* fcoe_ctlr_mtu_valid() - Check if a FCF's MTU is valid
* @fcf: The FCF to check
*
* Return non-zero if FCF fcoe_size has been validated.
*/
static inline int fcoe_ctlr_mtu_valid(const struct fcoe_fcf *fcf)
{
return (fcf->flags & FIP_FL_SOL) != 0;
}
/**
* fcoe_ctlr_fcf_usable() - Check if a FCF is usable
* @fcf: The FCF to check
*
* Return non-zero if the FCF is usable.
*/
static inline int fcoe_ctlr_fcf_usable(struct fcoe_fcf *fcf)
{
u16 flags = FIP_FL_SOL | FIP_FL_AVAIL;
return (fcf->flags & flags) == flags;
}
/**
* fcoe_ctlr_map_dest() - Set flag and OUI for mapping destination addresses
* @fip: The FCoE controller
*/
static void fcoe_ctlr_map_dest(struct fcoe_ctlr *fip)
{
if (fip->mode == FIP_MODE_VN2VN)
hton24(fip->dest_addr, FIP_VN_FC_MAP);
else
hton24(fip->dest_addr, FIP_DEF_FC_MAP);
hton24(fip->dest_addr + 3, 0);
fip->map_dest = 1;
}
/**
* fcoe_ctlr_init() - Initialize the FCoE Controller instance
* @fip: The FCoE controller to initialize
*/
void fcoe_ctlr_init(struct fcoe_ctlr *fip, enum fip_state mode)
{
fcoe_ctlr_set_state(fip, FIP_ST_LINK_WAIT);
fip->mode = mode;
fip->fip_resp = false;
INIT_LIST_HEAD(&fip->fcfs);
mutex_init(&fip->ctlr_mutex);
spin_lock_init(&fip->ctlr_lock);
fip->flogi_oxid = FC_XID_UNKNOWN;
treewide: setup_timer() -> timer_setup() This converts all remaining cases of the old setup_timer() API into using timer_setup(), where the callback argument is the structure already holding the struct timer_list. These should have no behavioral changes, since they just change which pointer is passed into the callback with the same available pointers after conversion. It handles the following examples, in addition to some other variations. Casting from unsigned long: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... setup_timer(&ptr->my_timer, my_callback, ptr); and forced object casts: void my_callback(struct something *ptr) { ... } ... setup_timer(&ptr->my_timer, my_callback, (unsigned long)ptr); become: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... timer_setup(&ptr->my_timer, my_callback, 0); Direct function assignments: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... ptr->my_timer.function = my_callback; have a temporary cast added, along with converting the args: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... ptr->my_timer.function = (TIMER_FUNC_TYPE)my_callback; And finally, callbacks without a data assignment: void my_callback(unsigned long data) { ... } ... setup_timer(&ptr->my_timer, my_callback, 0); have their argument renamed to verify they're unused during conversion: void my_callback(struct timer_list *unused) { ... } ... timer_setup(&ptr->my_timer, my_callback, 0); The conversion is done with the following Coccinelle script: spatch --very-quiet --all-includes --include-headers \ -I ./arch/x86/include -I ./arch/x86/include/generated \ -I ./include -I ./arch/x86/include/uapi \ -I ./arch/x86/include/generated/uapi -I ./include/uapi \ -I ./include/generated/uapi --include ./include/linux/kconfig.h \ --dir . \ --cocci-file ~/src/data/timer_setup.cocci @fix_address_of@ expression e; @@ setup_timer( -&(e) +&e , ...) // Update any raw setup_timer() usages that have a NULL callback, but // would otherwise match change_timer_function_usage, since the latter // will update all function assignments done in the face of a NULL // function initialization in setup_timer(). @change_timer_function_usage_NULL@ expression _E; identifier _timer; type _cast_data; @@ ( -setup_timer(&_E->_timer, NULL, _E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E->_timer, NULL, (_cast_data)_E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E._timer, NULL, &_E); +timer_setup(&_E._timer, NULL, 0); | -setup_timer(&_E._timer, NULL, (_cast_data)&_E); +timer_setup(&_E._timer, NULL, 0); ) @change_timer_function_usage@ expression _E; identifier _timer; struct timer_list _stl; identifier _callback; type _cast_func, _cast_data; @@ ( -setup_timer(&_E->_timer, _callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | _E->_timer@_stl.function = _callback; | _E->_timer@_stl.function = &_callback; | _E->_timer@_stl.function = (_cast_func)_callback; | _E->_timer@_stl.function = (_cast_func)&_callback; | _E._timer@_stl.function = _callback; | _E._timer@_stl.function = &_callback; | _E._timer@_stl.function = (_cast_func)_callback; | _E._timer@_stl.function = (_cast_func)&_callback; ) // callback(unsigned long arg) @change_callback_handle_cast depends on change_timer_function_usage@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; identifier _handle; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { ( ... when != _origarg _handletype *_handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg ) } // callback(unsigned long arg) without existing variable @change_callback_handle_cast_no_arg depends on change_timer_function_usage && !change_callback_handle_cast@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { + _handletype *_origarg = from_timer(_origarg, t, _timer); + ... when != _origarg - (_handletype *)_origarg + _origarg ... when != _origarg } // Avoid already converted callbacks. @match_callback_converted depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier t; @@ void _callback(struct timer_list *t) { ... } // callback(struct something *handle) @change_callback_handle_arg depends on change_timer_function_usage && !match_callback_converted && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; @@ void _callback( -_handletype *_handle +struct timer_list *t ) { + _handletype *_handle = from_timer(_handle, t, _timer); ... } // If change_callback_handle_arg ran on an empty function, remove // the added handler. @unchange_callback_handle_arg depends on change_timer_function_usage && change_callback_handle_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; identifier t; @@ void _callback(struct timer_list *t) { - _handletype *_handle = from_timer(_handle, t, _timer); } // We only want to refactor the setup_timer() data argument if we've found // the matching callback. This undoes changes in change_timer_function_usage. @unchange_timer_function_usage depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg && !change_callback_handle_arg@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type change_timer_function_usage._cast_data; @@ ( -timer_setup(&_E->_timer, _callback, 0); +setup_timer(&_E->_timer, _callback, (_cast_data)_E); | -timer_setup(&_E._timer, _callback, 0); +setup_timer(&_E._timer, _callback, (_cast_data)&_E); ) // If we fixed a callback from a .function assignment, fix the // assignment cast now. @change_timer_function_assignment depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_func; typedef TIMER_FUNC_TYPE; @@ ( _E->_timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -&_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)_callback; +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -&_callback; +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; ) // Sometimes timer functions are called directly. Replace matched args. @change_timer_function_calls depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression _E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_data; @@ _callback( ( -(_cast_data)_E +&_E->_timer | -(_cast_data)&_E +&_E._timer | -_E +&_E->_timer ) ) // If a timer has been configured without a data argument, it can be // converted without regard to the callback argument, since it is unused. @match_timer_function_unused_data@ expression _E; identifier _timer; identifier _callback; @@ ( -setup_timer(&_E->_timer, _callback, 0); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0L); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0UL); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0L); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0UL); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_timer, _callback, 0); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0L); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0UL); +timer_setup(&_timer, _callback, 0); | -setup_timer(_timer, _callback, 0); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0L); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0UL); +timer_setup(_timer, _callback, 0); ) @change_callback_unused_data depends on match_timer_function_unused_data@ identifier match_timer_function_unused_data._callback; type _origtype; identifier _origarg; @@ void _callback( -_origtype _origarg +struct timer_list *unused ) { ... when != _origarg } Signed-off-by: Kees Cook <keescook@chromium.org>
2017-10-17 05:43:17 +08:00
timer_setup(&fip->timer, fcoe_ctlr_timeout, 0);
INIT_WORK(&fip->timer_work, fcoe_ctlr_timer_work);
INIT_WORK(&fip->recv_work, fcoe_ctlr_recv_work);
skb_queue_head_init(&fip->fip_recv_list);
}
EXPORT_SYMBOL(fcoe_ctlr_init);
libfcoe: Make fcoe_sysfs optional / fix fnic NULL exception fnic doesn't use any of the create/destroy/enable/disable interfaces either from the (legacy) module paramaters or the (new) fcoe_sysfs interfaces. When fcoe_sysfs was introduced fnic wasn't changed since it wasn't using the interfaces. libfcoe incorrectly assumed that that all of its users were using fcoe_sysfs and when adding and deleting FCFs would assume the existance of a fcoe_ctlr_device. fnic was not allocating this structure because it doesn't care about the standard user interfaces (fnic starts on link only). If/When libfcoe tried to use the fcoe_ctlr_device's lock for the first time a NULL pointer exception would be triggered. Since fnic doesn't care about sysfs or user interfaces, the solution is to drop libfcoe's assumption that all drivers are using fcoe_sysfs. This patch accomplishes this by changing some of the structure relationships. We need a way to determine when a LLD is using fcoe_sysfs or not and we can do that by checking for the existance of the fcoe_ctlr_device. Prior to this patch, it was assumed that the fcoe_ctlr structure was allocated with the fcoe_ctlr_device and immediately followed it in memory. To reach the fcoe_ctlr_device we would simply go back in memory from the fcoe_ctlr to get the fcoe_ctlr_device. Since fnic doesn't allocate the fcoe_ctlr_device, we cannot keep that assumption. This patch adds a pointer from the fcoe_ctlr to the fcoe_ctlr_device. For bnx2fc and fcoe we will continue to allocate the two structures together, but then we'll set the ctlr->cdev pointer to point at the fcoe_ctlr_device. fnic will not change and will continue to allocate the fcoe_ctlr itself, and ctlr->cdev will remain NULL. When libfcoe adds fcoe_fcf's to the fcoe_ctlr it will check if ctlr->cdev is set and only if so will it continue to interact with fcoe_sysfs. Signed-off-by: Robert Love <robert.w.love@intel.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Tested-by: Hiral Patel <hiralpat@cisco.com>
2013-09-05 15:47:27 +08:00
/**
* fcoe_sysfs_fcf_add() - Add a fcoe_fcf{,_device} to a fcoe_ctlr{,_device}
* @new: The newly discovered FCF
*
* Called with fip->ctlr_mutex held
*/
static int fcoe_sysfs_fcf_add(struct fcoe_fcf *new)
{
struct fcoe_ctlr *fip = new->fip;
libfcoe: Make fcoe_sysfs optional / fix fnic NULL exception fnic doesn't use any of the create/destroy/enable/disable interfaces either from the (legacy) module paramaters or the (new) fcoe_sysfs interfaces. When fcoe_sysfs was introduced fnic wasn't changed since it wasn't using the interfaces. libfcoe incorrectly assumed that that all of its users were using fcoe_sysfs and when adding and deleting FCFs would assume the existance of a fcoe_ctlr_device. fnic was not allocating this structure because it doesn't care about the standard user interfaces (fnic starts on link only). If/When libfcoe tried to use the fcoe_ctlr_device's lock for the first time a NULL pointer exception would be triggered. Since fnic doesn't care about sysfs or user interfaces, the solution is to drop libfcoe's assumption that all drivers are using fcoe_sysfs. This patch accomplishes this by changing some of the structure relationships. We need a way to determine when a LLD is using fcoe_sysfs or not and we can do that by checking for the existance of the fcoe_ctlr_device. Prior to this patch, it was assumed that the fcoe_ctlr structure was allocated with the fcoe_ctlr_device and immediately followed it in memory. To reach the fcoe_ctlr_device we would simply go back in memory from the fcoe_ctlr to get the fcoe_ctlr_device. Since fnic doesn't allocate the fcoe_ctlr_device, we cannot keep that assumption. This patch adds a pointer from the fcoe_ctlr to the fcoe_ctlr_device. For bnx2fc and fcoe we will continue to allocate the two structures together, but then we'll set the ctlr->cdev pointer to point at the fcoe_ctlr_device. fnic will not change and will continue to allocate the fcoe_ctlr itself, and ctlr->cdev will remain NULL. When libfcoe adds fcoe_fcf's to the fcoe_ctlr it will check if ctlr->cdev is set and only if so will it continue to interact with fcoe_sysfs. Signed-off-by: Robert Love <robert.w.love@intel.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Tested-by: Hiral Patel <hiralpat@cisco.com>
2013-09-05 15:47:27 +08:00
struct fcoe_ctlr_device *ctlr_dev;
struct fcoe_fcf_device *temp, *fcf_dev;
int rc = -ENOMEM;
LIBFCOE_FIP_DBG(fip, "New FCF fab %16.16llx mac %pM\n",
new->fabric_name, new->fcf_mac);
temp = kzalloc(sizeof(*temp), GFP_KERNEL);
if (!temp)
goto out;
temp->fabric_name = new->fabric_name;
temp->switch_name = new->switch_name;
temp->fc_map = new->fc_map;
temp->vfid = new->vfid;
memcpy(temp->mac, new->fcf_mac, ETH_ALEN);
temp->priority = new->pri;
temp->fka_period = new->fka_period;
temp->selected = 0; /* default to unselected */
/*
libfcoe: Make fcoe_sysfs optional / fix fnic NULL exception fnic doesn't use any of the create/destroy/enable/disable interfaces either from the (legacy) module paramaters or the (new) fcoe_sysfs interfaces. When fcoe_sysfs was introduced fnic wasn't changed since it wasn't using the interfaces. libfcoe incorrectly assumed that that all of its users were using fcoe_sysfs and when adding and deleting FCFs would assume the existance of a fcoe_ctlr_device. fnic was not allocating this structure because it doesn't care about the standard user interfaces (fnic starts on link only). If/When libfcoe tried to use the fcoe_ctlr_device's lock for the first time a NULL pointer exception would be triggered. Since fnic doesn't care about sysfs or user interfaces, the solution is to drop libfcoe's assumption that all drivers are using fcoe_sysfs. This patch accomplishes this by changing some of the structure relationships. We need a way to determine when a LLD is using fcoe_sysfs or not and we can do that by checking for the existance of the fcoe_ctlr_device. Prior to this patch, it was assumed that the fcoe_ctlr structure was allocated with the fcoe_ctlr_device and immediately followed it in memory. To reach the fcoe_ctlr_device we would simply go back in memory from the fcoe_ctlr to get the fcoe_ctlr_device. Since fnic doesn't allocate the fcoe_ctlr_device, we cannot keep that assumption. This patch adds a pointer from the fcoe_ctlr to the fcoe_ctlr_device. For bnx2fc and fcoe we will continue to allocate the two structures together, but then we'll set the ctlr->cdev pointer to point at the fcoe_ctlr_device. fnic will not change and will continue to allocate the fcoe_ctlr itself, and ctlr->cdev will remain NULL. When libfcoe adds fcoe_fcf's to the fcoe_ctlr it will check if ctlr->cdev is set and only if so will it continue to interact with fcoe_sysfs. Signed-off-by: Robert Love <robert.w.love@intel.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Tested-by: Hiral Patel <hiralpat@cisco.com>
2013-09-05 15:47:27 +08:00
* If ctlr_dev doesn't exist then it means we're a libfcoe user
* who doesn't use fcoe_syfs and didn't allocate a fcoe_ctlr_device.
* fnic would be an example of a driver with this behavior. In this
* case we want to add the fcoe_fcf to the fcoe_ctlr list, but we
* don't want to make sysfs changes.
*/
libfcoe: Make fcoe_sysfs optional / fix fnic NULL exception fnic doesn't use any of the create/destroy/enable/disable interfaces either from the (legacy) module paramaters or the (new) fcoe_sysfs interfaces. When fcoe_sysfs was introduced fnic wasn't changed since it wasn't using the interfaces. libfcoe incorrectly assumed that that all of its users were using fcoe_sysfs and when adding and deleting FCFs would assume the existance of a fcoe_ctlr_device. fnic was not allocating this structure because it doesn't care about the standard user interfaces (fnic starts on link only). If/When libfcoe tried to use the fcoe_ctlr_device's lock for the first time a NULL pointer exception would be triggered. Since fnic doesn't care about sysfs or user interfaces, the solution is to drop libfcoe's assumption that all drivers are using fcoe_sysfs. This patch accomplishes this by changing some of the structure relationships. We need a way to determine when a LLD is using fcoe_sysfs or not and we can do that by checking for the existance of the fcoe_ctlr_device. Prior to this patch, it was assumed that the fcoe_ctlr structure was allocated with the fcoe_ctlr_device and immediately followed it in memory. To reach the fcoe_ctlr_device we would simply go back in memory from the fcoe_ctlr to get the fcoe_ctlr_device. Since fnic doesn't allocate the fcoe_ctlr_device, we cannot keep that assumption. This patch adds a pointer from the fcoe_ctlr to the fcoe_ctlr_device. For bnx2fc and fcoe we will continue to allocate the two structures together, but then we'll set the ctlr->cdev pointer to point at the fcoe_ctlr_device. fnic will not change and will continue to allocate the fcoe_ctlr itself, and ctlr->cdev will remain NULL. When libfcoe adds fcoe_fcf's to the fcoe_ctlr it will check if ctlr->cdev is set and only if so will it continue to interact with fcoe_sysfs. Signed-off-by: Robert Love <robert.w.love@intel.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Tested-by: Hiral Patel <hiralpat@cisco.com>
2013-09-05 15:47:27 +08:00
ctlr_dev = fcoe_ctlr_to_ctlr_dev(fip);
if (ctlr_dev) {
mutex_lock(&ctlr_dev->lock);
fcf_dev = fcoe_fcf_device_add(ctlr_dev, temp);
if (unlikely(!fcf_dev)) {
rc = -ENOMEM;
mutex_unlock(&ctlr_dev->lock);
libfcoe: Make fcoe_sysfs optional / fix fnic NULL exception fnic doesn't use any of the create/destroy/enable/disable interfaces either from the (legacy) module paramaters or the (new) fcoe_sysfs interfaces. When fcoe_sysfs was introduced fnic wasn't changed since it wasn't using the interfaces. libfcoe incorrectly assumed that that all of its users were using fcoe_sysfs and when adding and deleting FCFs would assume the existance of a fcoe_ctlr_device. fnic was not allocating this structure because it doesn't care about the standard user interfaces (fnic starts on link only). If/When libfcoe tried to use the fcoe_ctlr_device's lock for the first time a NULL pointer exception would be triggered. Since fnic doesn't care about sysfs or user interfaces, the solution is to drop libfcoe's assumption that all drivers are using fcoe_sysfs. This patch accomplishes this by changing some of the structure relationships. We need a way to determine when a LLD is using fcoe_sysfs or not and we can do that by checking for the existance of the fcoe_ctlr_device. Prior to this patch, it was assumed that the fcoe_ctlr structure was allocated with the fcoe_ctlr_device and immediately followed it in memory. To reach the fcoe_ctlr_device we would simply go back in memory from the fcoe_ctlr to get the fcoe_ctlr_device. Since fnic doesn't allocate the fcoe_ctlr_device, we cannot keep that assumption. This patch adds a pointer from the fcoe_ctlr to the fcoe_ctlr_device. For bnx2fc and fcoe we will continue to allocate the two structures together, but then we'll set the ctlr->cdev pointer to point at the fcoe_ctlr_device. fnic will not change and will continue to allocate the fcoe_ctlr itself, and ctlr->cdev will remain NULL. When libfcoe adds fcoe_fcf's to the fcoe_ctlr it will check if ctlr->cdev is set and only if so will it continue to interact with fcoe_sysfs. Signed-off-by: Robert Love <robert.w.love@intel.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Tested-by: Hiral Patel <hiralpat@cisco.com>
2013-09-05 15:47:27 +08:00
goto out;
}
/*
* The fcoe_sysfs layer can return a CONNECTED fcf that
* has a priv (fcf was never deleted) or a CONNECTED fcf
* that doesn't have a priv (fcf was deleted). However,
* libfcoe will always delete FCFs before trying to add
* them. This is ensured because both recv_adv and
* age_fcfs are protected by the the fcoe_ctlr's mutex.
* This means that we should never get a FCF with a
* non-NULL priv pointer.
*/
BUG_ON(fcf_dev->priv);
fcf_dev->priv = new;
new->fcf_dev = fcf_dev;
mutex_unlock(&ctlr_dev->lock);
}
list_add(&new->list, &fip->fcfs);
fip->fcf_count++;
rc = 0;
out:
kfree(temp);
return rc;
}
libfcoe: Make fcoe_sysfs optional / fix fnic NULL exception fnic doesn't use any of the create/destroy/enable/disable interfaces either from the (legacy) module paramaters or the (new) fcoe_sysfs interfaces. When fcoe_sysfs was introduced fnic wasn't changed since it wasn't using the interfaces. libfcoe incorrectly assumed that that all of its users were using fcoe_sysfs and when adding and deleting FCFs would assume the existance of a fcoe_ctlr_device. fnic was not allocating this structure because it doesn't care about the standard user interfaces (fnic starts on link only). If/When libfcoe tried to use the fcoe_ctlr_device's lock for the first time a NULL pointer exception would be triggered. Since fnic doesn't care about sysfs or user interfaces, the solution is to drop libfcoe's assumption that all drivers are using fcoe_sysfs. This patch accomplishes this by changing some of the structure relationships. We need a way to determine when a LLD is using fcoe_sysfs or not and we can do that by checking for the existance of the fcoe_ctlr_device. Prior to this patch, it was assumed that the fcoe_ctlr structure was allocated with the fcoe_ctlr_device and immediately followed it in memory. To reach the fcoe_ctlr_device we would simply go back in memory from the fcoe_ctlr to get the fcoe_ctlr_device. Since fnic doesn't allocate the fcoe_ctlr_device, we cannot keep that assumption. This patch adds a pointer from the fcoe_ctlr to the fcoe_ctlr_device. For bnx2fc and fcoe we will continue to allocate the two structures together, but then we'll set the ctlr->cdev pointer to point at the fcoe_ctlr_device. fnic will not change and will continue to allocate the fcoe_ctlr itself, and ctlr->cdev will remain NULL. When libfcoe adds fcoe_fcf's to the fcoe_ctlr it will check if ctlr->cdev is set and only if so will it continue to interact with fcoe_sysfs. Signed-off-by: Robert Love <robert.w.love@intel.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Tested-by: Hiral Patel <hiralpat@cisco.com>
2013-09-05 15:47:27 +08:00
/**
* fcoe_sysfs_fcf_del() - Remove a fcoe_fcf{,_device} to a fcoe_ctlr{,_device}
* @new: The FCF to be removed
*
* Called with fip->ctlr_mutex held
*/
static void fcoe_sysfs_fcf_del(struct fcoe_fcf *new)
{
struct fcoe_ctlr *fip = new->fip;
libfcoe: Make fcoe_sysfs optional / fix fnic NULL exception fnic doesn't use any of the create/destroy/enable/disable interfaces either from the (legacy) module paramaters or the (new) fcoe_sysfs interfaces. When fcoe_sysfs was introduced fnic wasn't changed since it wasn't using the interfaces. libfcoe incorrectly assumed that that all of its users were using fcoe_sysfs and when adding and deleting FCFs would assume the existance of a fcoe_ctlr_device. fnic was not allocating this structure because it doesn't care about the standard user interfaces (fnic starts on link only). If/When libfcoe tried to use the fcoe_ctlr_device's lock for the first time a NULL pointer exception would be triggered. Since fnic doesn't care about sysfs or user interfaces, the solution is to drop libfcoe's assumption that all drivers are using fcoe_sysfs. This patch accomplishes this by changing some of the structure relationships. We need a way to determine when a LLD is using fcoe_sysfs or not and we can do that by checking for the existance of the fcoe_ctlr_device. Prior to this patch, it was assumed that the fcoe_ctlr structure was allocated with the fcoe_ctlr_device and immediately followed it in memory. To reach the fcoe_ctlr_device we would simply go back in memory from the fcoe_ctlr to get the fcoe_ctlr_device. Since fnic doesn't allocate the fcoe_ctlr_device, we cannot keep that assumption. This patch adds a pointer from the fcoe_ctlr to the fcoe_ctlr_device. For bnx2fc and fcoe we will continue to allocate the two structures together, but then we'll set the ctlr->cdev pointer to point at the fcoe_ctlr_device. fnic will not change and will continue to allocate the fcoe_ctlr itself, and ctlr->cdev will remain NULL. When libfcoe adds fcoe_fcf's to the fcoe_ctlr it will check if ctlr->cdev is set and only if so will it continue to interact with fcoe_sysfs. Signed-off-by: Robert Love <robert.w.love@intel.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Tested-by: Hiral Patel <hiralpat@cisco.com>
2013-09-05 15:47:27 +08:00
struct fcoe_ctlr_device *cdev;
struct fcoe_fcf_device *fcf_dev;
list_del(&new->list);
fip->fcf_count--;
libfcoe: Make fcoe_sysfs optional / fix fnic NULL exception fnic doesn't use any of the create/destroy/enable/disable interfaces either from the (legacy) module paramaters or the (new) fcoe_sysfs interfaces. When fcoe_sysfs was introduced fnic wasn't changed since it wasn't using the interfaces. libfcoe incorrectly assumed that that all of its users were using fcoe_sysfs and when adding and deleting FCFs would assume the existance of a fcoe_ctlr_device. fnic was not allocating this structure because it doesn't care about the standard user interfaces (fnic starts on link only). If/When libfcoe tried to use the fcoe_ctlr_device's lock for the first time a NULL pointer exception would be triggered. Since fnic doesn't care about sysfs or user interfaces, the solution is to drop libfcoe's assumption that all drivers are using fcoe_sysfs. This patch accomplishes this by changing some of the structure relationships. We need a way to determine when a LLD is using fcoe_sysfs or not and we can do that by checking for the existance of the fcoe_ctlr_device. Prior to this patch, it was assumed that the fcoe_ctlr structure was allocated with the fcoe_ctlr_device and immediately followed it in memory. To reach the fcoe_ctlr_device we would simply go back in memory from the fcoe_ctlr to get the fcoe_ctlr_device. Since fnic doesn't allocate the fcoe_ctlr_device, we cannot keep that assumption. This patch adds a pointer from the fcoe_ctlr to the fcoe_ctlr_device. For bnx2fc and fcoe we will continue to allocate the two structures together, but then we'll set the ctlr->cdev pointer to point at the fcoe_ctlr_device. fnic will not change and will continue to allocate the fcoe_ctlr itself, and ctlr->cdev will remain NULL. When libfcoe adds fcoe_fcf's to the fcoe_ctlr it will check if ctlr->cdev is set and only if so will it continue to interact with fcoe_sysfs. Signed-off-by: Robert Love <robert.w.love@intel.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Tested-by: Hiral Patel <hiralpat@cisco.com>
2013-09-05 15:47:27 +08:00
/*
* If ctlr_dev doesn't exist then it means we're a libfcoe user
* who doesn't use fcoe_syfs and didn't allocate a fcoe_ctlr_device
* or a fcoe_fcf_device.
*
* fnic would be an example of a driver with this behavior. In this
* case we want to remove the fcoe_fcf from the fcoe_ctlr list (above),
* but we don't want to make sysfs changes.
*/
cdev = fcoe_ctlr_to_ctlr_dev(fip);
if (cdev) {
mutex_lock(&cdev->lock);
fcf_dev = fcoe_fcf_to_fcf_dev(new);
WARN_ON(!fcf_dev);
new->fcf_dev = NULL;
fcoe_fcf_device_delete(fcf_dev);
kfree(new);
mutex_unlock(&cdev->lock);
}
}
/**
* fcoe_ctlr_reset_fcfs() - Reset and free all FCFs for a controller
* @fip: The FCoE controller whose FCFs are to be reset
*
* Called with &fcoe_ctlr lock held.
*/
static void fcoe_ctlr_reset_fcfs(struct fcoe_ctlr *fip)
{
struct fcoe_fcf *fcf;
struct fcoe_fcf *next;
fip->sel_fcf = NULL;
list_for_each_entry_safe(fcf, next, &fip->fcfs, list) {
fcoe_sysfs_fcf_del(fcf);
}
WARN_ON(fip->fcf_count);
fip->sel_time = 0;
}
/**
* fcoe_ctlr_destroy() - Disable and tear down a FCoE controller
* @fip: The FCoE controller to tear down
*
* This is called by FCoE drivers before freeing the &fcoe_ctlr.
*
* The receive handler will have been deleted before this to guarantee
* that no more recv_work will be scheduled.
*
* The timer routine will simply return once we set FIP_ST_DISABLED.
* This guarantees that no further timeouts or work will be scheduled.
*/
void fcoe_ctlr_destroy(struct fcoe_ctlr *fip)
{
cancel_work_sync(&fip->recv_work);
skb_queue_purge(&fip->fip_recv_list);
mutex_lock(&fip->ctlr_mutex);
fcoe_ctlr_set_state(fip, FIP_ST_DISABLED);
fcoe_ctlr_reset_fcfs(fip);
mutex_unlock(&fip->ctlr_mutex);
del_timer_sync(&fip->timer);
cancel_work_sync(&fip->timer_work);
}
EXPORT_SYMBOL(fcoe_ctlr_destroy);
/**
* fcoe_ctlr_announce() - announce new FCF selection
* @fip: The FCoE controller
*
* Also sets the destination MAC for FCoE and control packets
*
* Called with neither ctlr_mutex nor ctlr_lock held.
*/
static void fcoe_ctlr_announce(struct fcoe_ctlr *fip)
{
struct fcoe_fcf *sel;
struct fcoe_fcf *fcf;
mutex_lock(&fip->ctlr_mutex);
spin_lock_bh(&fip->ctlr_lock);
kfree_skb(fip->flogi_req);
fip->flogi_req = NULL;
list_for_each_entry(fcf, &fip->fcfs, list)
fcf->flogi_sent = 0;
spin_unlock_bh(&fip->ctlr_lock);
sel = fip->sel_fcf;
if (sel && ether_addr_equal(sel->fcf_mac, fip->dest_addr))
goto unlock;
if (!is_zero_ether_addr(fip->dest_addr)) {
printk(KERN_NOTICE "libfcoe: host%d: "
"FIP Fibre-Channel Forwarder MAC %pM deselected\n",
fip->lp->host->host_no, fip->dest_addr);
memset(fip->dest_addr, 0, ETH_ALEN);
}
if (sel) {
printk(KERN_INFO "libfcoe: host%d: FIP selected "
"Fibre-Channel Forwarder MAC %pM\n",
fip->lp->host->host_no, sel->fcf_mac);
memcpy(fip->dest_addr, sel->fcoe_mac, ETH_ALEN);
fip->map_dest = 0;
}
unlock:
mutex_unlock(&fip->ctlr_mutex);
}
/**
* fcoe_ctlr_fcoe_size() - Return the maximum FCoE size required for VN_Port
* @fip: The FCoE controller to get the maximum FCoE size from
*
* Returns the maximum packet size including the FCoE header and trailer,
* but not including any Ethernet or VLAN headers.
*/
static inline u32 fcoe_ctlr_fcoe_size(struct fcoe_ctlr *fip)
{
/*
* Determine the max FCoE frame size allowed, including
* FCoE header and trailer.
* Note: lp->mfs is currently the payload size, not the frame size.
*/
return fip->lp->mfs + sizeof(struct fc_frame_header) +
sizeof(struct fcoe_hdr) + sizeof(struct fcoe_crc_eof);
}
/**
* fcoe_ctlr_solicit() - Send a FIP solicitation
* @fip: The FCoE controller to send the solicitation on
* @fcf: The destination FCF (if NULL, a multicast solicitation is sent)
*/
static void fcoe_ctlr_solicit(struct fcoe_ctlr *fip, struct fcoe_fcf *fcf)
{
struct sk_buff *skb;
struct fip_sol {
struct ethhdr eth;
struct fip_header fip;
struct {
struct fip_mac_desc mac;
struct fip_wwn_desc wwnn;
struct fip_size_desc size;
} __packed desc;
} __packed * sol;
u32 fcoe_size;
skb = dev_alloc_skb(sizeof(*sol));
if (!skb)
return;
sol = (struct fip_sol *)skb->data;
memset(sol, 0, sizeof(*sol));
memcpy(sol->eth.h_dest, fcf ? fcf->fcf_mac : fcoe_all_fcfs, ETH_ALEN);
memcpy(sol->eth.h_source, fip->ctl_src_addr, ETH_ALEN);
sol->eth.h_proto = htons(ETH_P_FIP);
sol->fip.fip_ver = FIP_VER_ENCAPS(FIP_VER);
sol->fip.fip_op = htons(FIP_OP_DISC);
sol->fip.fip_subcode = FIP_SC_SOL;
sol->fip.fip_dl_len = htons(sizeof(sol->desc) / FIP_BPW);
sol->fip.fip_flags = htons(FIP_FL_FPMA);
if (fip->spma)
sol->fip.fip_flags |= htons(FIP_FL_SPMA);
sol->desc.mac.fd_desc.fip_dtype = FIP_DT_MAC;
sol->desc.mac.fd_desc.fip_dlen = sizeof(sol->desc.mac) / FIP_BPW;
memcpy(sol->desc.mac.fd_mac, fip->ctl_src_addr, ETH_ALEN);
sol->desc.wwnn.fd_desc.fip_dtype = FIP_DT_NAME;
sol->desc.wwnn.fd_desc.fip_dlen = sizeof(sol->desc.wwnn) / FIP_BPW;
put_unaligned_be64(fip->lp->wwnn, &sol->desc.wwnn.fd_wwn);
fcoe_size = fcoe_ctlr_fcoe_size(fip);
sol->desc.size.fd_desc.fip_dtype = FIP_DT_FCOE_SIZE;
sol->desc.size.fd_desc.fip_dlen = sizeof(sol->desc.size) / FIP_BPW;
sol->desc.size.fd_size = htons(fcoe_size);
skb_put(skb, sizeof(*sol));
skb->protocol = htons(ETH_P_FIP);
skb->priority = fip->priority;
skb_reset_mac_header(skb);
skb_reset_network_header(skb);
fip->send(fip, skb);
if (!fcf)
fip->sol_time = jiffies;
}
/**
* fcoe_ctlr_link_up() - Start FCoE controller
* @fip: The FCoE controller to start
*
* Called from the LLD when the network link is ready.
*/
void fcoe_ctlr_link_up(struct fcoe_ctlr *fip)
{
mutex_lock(&fip->ctlr_mutex);
if (fip->state == FIP_ST_NON_FIP || fip->state == FIP_ST_AUTO) {
mutex_unlock(&fip->ctlr_mutex);
fc_linkup(fip->lp);
} else if (fip->state == FIP_ST_LINK_WAIT) {
fcoe_ctlr_set_state(fip, fip->mode);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
switch (fip->mode) {
default:
LIBFCOE_FIP_DBG(fip, "invalid mode %d\n", fip->mode);
/* fall-through */
case FIP_MODE_AUTO:
LIBFCOE_FIP_DBG(fip, "%s", "setting AUTO mode.\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
/* fall-through */
case FIP_MODE_FABRIC:
case FIP_MODE_NON_FIP:
mutex_unlock(&fip->ctlr_mutex);
fc_linkup(fip->lp);
fcoe_ctlr_solicit(fip, NULL);
break;
case FIP_MODE_VN2VN:
fcoe_ctlr_vn_start(fip);
mutex_unlock(&fip->ctlr_mutex);
fc_linkup(fip->lp);
break;
}
} else
mutex_unlock(&fip->ctlr_mutex);
}
EXPORT_SYMBOL(fcoe_ctlr_link_up);
/**
* fcoe_ctlr_reset() - Reset a FCoE controller
* @fip: The FCoE controller to reset
*/
static void fcoe_ctlr_reset(struct fcoe_ctlr *fip)
{
fcoe_ctlr_reset_fcfs(fip);
del_timer(&fip->timer);
fip->ctlr_ka_time = 0;
fip->port_ka_time = 0;
fip->sol_time = 0;
fip->flogi_oxid = FC_XID_UNKNOWN;
fcoe_ctlr_map_dest(fip);
}
/**
* fcoe_ctlr_link_down() - Stop a FCoE controller
* @fip: The FCoE controller to be stopped
*
* Returns non-zero if the link was up and now isn't.
*
* Called from the LLD when the network link is not ready.
* There may be multiple calls while the link is down.
*/
int fcoe_ctlr_link_down(struct fcoe_ctlr *fip)
{
int link_dropped;
LIBFCOE_FIP_DBG(fip, "link down.\n");
mutex_lock(&fip->ctlr_mutex);
fcoe_ctlr_reset(fip);
link_dropped = fip->state != FIP_ST_LINK_WAIT;
fcoe_ctlr_set_state(fip, FIP_ST_LINK_WAIT);
mutex_unlock(&fip->ctlr_mutex);
if (link_dropped)
fc_linkdown(fip->lp);
return link_dropped;
}
EXPORT_SYMBOL(fcoe_ctlr_link_down);
/**
* fcoe_ctlr_send_keep_alive() - Send a keep-alive to the selected FCF
* @fip: The FCoE controller to send the FKA on
* @lport: libfc fc_lport to send from
* @ports: 0 for controller keep-alive, 1 for port keep-alive
* @sa: The source MAC address
*
* A controller keep-alive is sent every fka_period (typically 8 seconds).
* The source MAC is the native MAC address.
*
* A port keep-alive is sent every 90 seconds while logged in.
* The source MAC is the assigned mapped source address.
* The destination is the FCF's F-port.
*/
static void fcoe_ctlr_send_keep_alive(struct fcoe_ctlr *fip,
struct fc_lport *lport,
int ports, u8 *sa)
{
struct sk_buff *skb;
struct fip_kal {
struct ethhdr eth;
struct fip_header fip;
struct fip_mac_desc mac;
} __packed * kal;
struct fip_vn_desc *vn;
u32 len;
struct fc_lport *lp;
struct fcoe_fcf *fcf;
fcf = fip->sel_fcf;
lp = fip->lp;
if (!fcf || (ports && !lp->port_id))
return;
len = sizeof(*kal) + ports * sizeof(*vn);
skb = dev_alloc_skb(len);
if (!skb)
return;
kal = (struct fip_kal *)skb->data;
memset(kal, 0, len);
memcpy(kal->eth.h_dest, fcf->fcf_mac, ETH_ALEN);
memcpy(kal->eth.h_source, sa, ETH_ALEN);
kal->eth.h_proto = htons(ETH_P_FIP);
kal->fip.fip_ver = FIP_VER_ENCAPS(FIP_VER);
kal->fip.fip_op = htons(FIP_OP_CTRL);
kal->fip.fip_subcode = FIP_SC_KEEP_ALIVE;
kal->fip.fip_dl_len = htons((sizeof(kal->mac) +
ports * sizeof(*vn)) / FIP_BPW);
kal->fip.fip_flags = htons(FIP_FL_FPMA);
if (fip->spma)
kal->fip.fip_flags |= htons(FIP_FL_SPMA);
kal->mac.fd_desc.fip_dtype = FIP_DT_MAC;
kal->mac.fd_desc.fip_dlen = sizeof(kal->mac) / FIP_BPW;
memcpy(kal->mac.fd_mac, fip->ctl_src_addr, ETH_ALEN);
if (ports) {
vn = (struct fip_vn_desc *)(kal + 1);
vn->fd_desc.fip_dtype = FIP_DT_VN_ID;
vn->fd_desc.fip_dlen = sizeof(*vn) / FIP_BPW;
memcpy(vn->fd_mac, fip->get_src_addr(lport), ETH_ALEN);
hton24(vn->fd_fc_id, lport->port_id);
put_unaligned_be64(lport->wwpn, &vn->fd_wwpn);
}
skb_put(skb, len);
skb->protocol = htons(ETH_P_FIP);
skb->priority = fip->priority;
skb_reset_mac_header(skb);
skb_reset_network_header(skb);
fip->send(fip, skb);
}
/**
* fcoe_ctlr_encaps() - Encapsulate an ELS frame for FIP, without sending it
* @fip: The FCoE controller for the ELS frame
* @dtype: The FIP descriptor type for the frame
* @skb: The FCoE ELS frame including FC header but no FCoE headers
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
* @d_id: The destination port ID.
*
* Returns non-zero error code on failure.
*
* The caller must check that the length is a multiple of 4.
*
* The @skb must have enough headroom (28 bytes) and tailroom (8 bytes).
* Headroom includes the FIP encapsulation description, FIP header, and
* Ethernet header. The tailroom is for the FIP MAC descriptor.
*/
static int fcoe_ctlr_encaps(struct fcoe_ctlr *fip, struct fc_lport *lport,
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
u8 dtype, struct sk_buff *skb, u32 d_id)
{
struct fip_encaps_head {
struct ethhdr eth;
struct fip_header fip;
struct fip_encaps encaps;
} __packed * cap;
struct fc_frame_header *fh;
struct fip_mac_desc *mac;
struct fcoe_fcf *fcf;
size_t dlen;
u16 fip_flags;
u8 op;
fh = (struct fc_frame_header *)skb->data;
op = *(u8 *)(fh + 1);
dlen = sizeof(struct fip_encaps) + skb->len; /* len before push */
cap = skb_push(skb, sizeof(*cap));
memset(cap, 0, sizeof(*cap));
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (lport->point_to_multipoint) {
if (fcoe_ctlr_vn_lookup(fip, d_id, cap->eth.h_dest))
return -ENODEV;
fip_flags = 0;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
} else {
fcf = fip->sel_fcf;
if (!fcf)
return -ENODEV;
fip_flags = fcf->flags;
fip_flags &= fip->spma ? FIP_FL_SPMA | FIP_FL_FPMA :
FIP_FL_FPMA;
if (!fip_flags)
return -ENODEV;
memcpy(cap->eth.h_dest, fcf->fcf_mac, ETH_ALEN);
}
memcpy(cap->eth.h_source, fip->ctl_src_addr, ETH_ALEN);
cap->eth.h_proto = htons(ETH_P_FIP);
cap->fip.fip_ver = FIP_VER_ENCAPS(FIP_VER);
cap->fip.fip_op = htons(FIP_OP_LS);
if (op == ELS_LS_ACC || op == ELS_LS_RJT)
cap->fip.fip_subcode = FIP_SC_REP;
else
cap->fip.fip_subcode = FIP_SC_REQ;
cap->fip.fip_flags = htons(fip_flags);
cap->encaps.fd_desc.fip_dtype = dtype;
cap->encaps.fd_desc.fip_dlen = dlen / FIP_BPW;
if (op != ELS_LS_RJT) {
dlen += sizeof(*mac);
mac = skb_put_zero(skb, sizeof(*mac));
mac->fd_desc.fip_dtype = FIP_DT_MAC;
mac->fd_desc.fip_dlen = sizeof(*mac) / FIP_BPW;
if (dtype != FIP_DT_FLOGI && dtype != FIP_DT_FDISC) {
memcpy(mac->fd_mac, fip->get_src_addr(lport), ETH_ALEN);
} else if (fip->mode == FIP_MODE_VN2VN) {
hton24(mac->fd_mac, FIP_VN_FC_MAP);
hton24(mac->fd_mac + 3, fip->port_id);
} else if (fip_flags & FIP_FL_SPMA) {
LIBFCOE_FIP_DBG(fip, "FLOGI/FDISC sent with SPMA\n");
memcpy(mac->fd_mac, fip->ctl_src_addr, ETH_ALEN);
} else {
LIBFCOE_FIP_DBG(fip, "FLOGI/FDISC sent with FPMA\n");
/* FPMA only FLOGI. Must leave the MAC desc zeroed. */
}
}
cap->fip.fip_dl_len = htons(dlen / FIP_BPW);
skb->protocol = htons(ETH_P_FIP);
skb->priority = fip->priority;
skb_reset_mac_header(skb);
skb_reset_network_header(skb);
return 0;
}
/**
* fcoe_ctlr_els_send() - Send an ELS frame encapsulated by FIP if appropriate.
* @fip: FCoE controller.
* @lport: libfc fc_lport to send from
* @skb: FCoE ELS frame including FC header but no FCoE headers.
*
* Returns a non-zero error code if the frame should not be sent.
* Returns zero if the caller should send the frame with FCoE encapsulation.
*
* The caller must check that the length is a multiple of 4.
* The SKB must have enough headroom (28 bytes) and tailroom (8 bytes).
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
* The the skb must also be an fc_frame.
*
* This is called from the lower-level driver with spinlocks held,
* so we must not take a mutex here.
*/
int fcoe_ctlr_els_send(struct fcoe_ctlr *fip, struct fc_lport *lport,
struct sk_buff *skb)
{
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
struct fc_frame *fp;
struct fc_frame_header *fh;
u16 old_xid;
u8 op;
u8 mac[ETH_ALEN];
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fp = container_of(skb, struct fc_frame, skb);
fh = (struct fc_frame_header *)skb->data;
op = *(u8 *)(fh + 1);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (op == ELS_FLOGI && fip->mode != FIP_MODE_VN2VN) {
old_xid = fip->flogi_oxid;
fip->flogi_oxid = ntohs(fh->fh_ox_id);
if (fip->state == FIP_ST_AUTO) {
if (old_xid == FC_XID_UNKNOWN)
fip->flogi_count = 0;
fip->flogi_count++;
if (fip->flogi_count < 3)
goto drop;
fcoe_ctlr_map_dest(fip);
return 0;
}
if (fip->state == FIP_ST_NON_FIP)
fcoe_ctlr_map_dest(fip);
}
if (fip->state == FIP_ST_NON_FIP)
return 0;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (!fip->sel_fcf && fip->mode != FIP_MODE_VN2VN)
goto drop;
switch (op) {
case ELS_FLOGI:
op = FIP_DT_FLOGI;
if (fip->mode == FIP_MODE_VN2VN)
break;
spin_lock_bh(&fip->ctlr_lock);
kfree_skb(fip->flogi_req);
fip->flogi_req = skb;
fip->flogi_req_send = 1;
spin_unlock_bh(&fip->ctlr_lock);
schedule_work(&fip->timer_work);
return -EINPROGRESS;
case ELS_FDISC:
if (ntoh24(fh->fh_s_id))
return 0;
op = FIP_DT_FDISC;
break;
case ELS_LOGO:
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (fip->mode == FIP_MODE_VN2VN) {
if (fip->state != FIP_ST_VNMP_UP)
return -EINVAL;
if (ntoh24(fh->fh_d_id) == FC_FID_FLOGI)
return -EINVAL;
} else {
if (fip->state != FIP_ST_ENABLED)
return 0;
if (ntoh24(fh->fh_d_id) != FC_FID_FLOGI)
return 0;
}
op = FIP_DT_LOGO;
break;
case ELS_LS_ACC:
/*
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
* If non-FIP, we may have gotten an SID by accepting an FLOGI
* from a point-to-point connection. Switch to using
* the source mac based on the SID. The destination
* MAC in this case would have been set by receiving the
* FLOGI.
*/
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (fip->state == FIP_ST_NON_FIP) {
if (fip->flogi_oxid == FC_XID_UNKNOWN)
return 0;
fip->flogi_oxid = FC_XID_UNKNOWN;
fc_fcoe_set_mac(mac, fh->fh_d_id);
fip->update_mac(lport, mac);
}
/* fall through */
case ELS_LS_RJT:
op = fr_encaps(fp);
if (op)
break;
return 0;
default:
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (fip->state != FIP_ST_ENABLED &&
fip->state != FIP_ST_VNMP_UP)
goto drop;
return 0;
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
LIBFCOE_FIP_DBG(fip, "els_send op %u d_id %x\n",
op, ntoh24(fh->fh_d_id));
if (fcoe_ctlr_encaps(fip, lport, op, skb, ntoh24(fh->fh_d_id)))
goto drop;
fip->send(fip, skb);
return -EINPROGRESS;
drop:
kfree_skb(skb);
LIBFCOE_FIP_DBG(fip, "drop els_send op %u d_id %x\n",
op, ntoh24(fh->fh_d_id));
return -EINVAL;
}
EXPORT_SYMBOL(fcoe_ctlr_els_send);
/**
* fcoe_ctlr_age_fcfs() - Reset and free all old FCFs for a controller
* @fip: The FCoE controller to free FCFs on
*
* Called with lock held and preemption disabled.
*
* An FCF is considered old if we have missed two advertisements.
* That is, there have been no valid advertisement from it for 2.5
* times its keep-alive period.
*
* In addition, determine the time when an FCF selection can occur.
*
* Also, increment the MissDiscAdvCount when no advertisement is received
* for the corresponding FCF for 1.5 * FKA_ADV_PERIOD (FC-BB-5 LESB).
*
* Returns the time in jiffies for the next call.
*/
static unsigned long fcoe_ctlr_age_fcfs(struct fcoe_ctlr *fip)
{
struct fcoe_fcf *fcf;
struct fcoe_fcf *next;
unsigned long next_timer = jiffies + msecs_to_jiffies(FIP_VN_KA_PERIOD);
unsigned long deadline;
unsigned long sel_time = 0;
struct list_head del_list;
struct fc_stats *stats;
INIT_LIST_HEAD(&del_list);
stats = per_cpu_ptr(fip->lp->stats, get_cpu());
list_for_each_entry_safe(fcf, next, &fip->fcfs, list) {
deadline = fcf->time + fcf->fka_period + fcf->fka_period / 2;
if (fip->sel_fcf == fcf) {
if (time_after(jiffies, deadline)) {
stats->MissDiscAdvCount++;
printk(KERN_INFO "libfcoe: host%d: "
"Missing Discovery Advertisement "
"for fab %16.16llx count %lld\n",
fip->lp->host->host_no, fcf->fabric_name,
stats->MissDiscAdvCount);
} else if (time_after(next_timer, deadline))
next_timer = deadline;
}
deadline += fcf->fka_period;
if (time_after_eq(jiffies, deadline)) {
if (fip->sel_fcf == fcf)
fip->sel_fcf = NULL;
/*
* Move to delete list so we can call
* fcoe_sysfs_fcf_del (which can sleep)
* after the put_cpu().
*/
list_del(&fcf->list);
list_add(&fcf->list, &del_list);
stats->VLinkFailureCount++;
} else {
if (time_after(next_timer, deadline))
next_timer = deadline;
if (fcoe_ctlr_mtu_valid(fcf) &&
(!sel_time || time_before(sel_time, fcf->time)))
sel_time = fcf->time;
}
}
put_cpu();
list_for_each_entry_safe(fcf, next, &del_list, list) {
/* Removes fcf from current list */
fcoe_sysfs_fcf_del(fcf);
}
if (sel_time && !fip->sel_fcf && !fip->sel_time) {
sel_time += msecs_to_jiffies(FCOE_CTLR_START_DELAY);
fip->sel_time = sel_time;
}
return next_timer;
}
/**
* fcoe_ctlr_parse_adv() - Decode a FIP advertisement into a new FCF entry
* @fip: The FCoE controller receiving the advertisement
* @skb: The received FIP advertisement frame
* @fcf: The resulting FCF entry
*
* Returns zero on a valid parsed advertisement,
* otherwise returns non zero value.
*/
static int fcoe_ctlr_parse_adv(struct fcoe_ctlr *fip,
struct sk_buff *skb, struct fcoe_fcf *fcf)
{
struct fip_header *fiph;
struct fip_desc *desc = NULL;
struct fip_wwn_desc *wwn;
struct fip_fab_desc *fab;
struct fip_fka_desc *fka;
unsigned long t;
size_t rlen;
size_t dlen;
u32 desc_mask;
memset(fcf, 0, sizeof(*fcf));
fcf->fka_period = msecs_to_jiffies(FCOE_CTLR_DEF_FKA);
fiph = (struct fip_header *)skb->data;
fcf->flags = ntohs(fiph->fip_flags);
/*
* mask of required descriptors. validating each one clears its bit.
*/
desc_mask = BIT(FIP_DT_PRI) | BIT(FIP_DT_MAC) | BIT(FIP_DT_NAME) |
BIT(FIP_DT_FAB) | BIT(FIP_DT_FKA);
rlen = ntohs(fiph->fip_dl_len) * 4;
if (rlen + sizeof(*fiph) > skb->len)
return -EINVAL;
desc = (struct fip_desc *)(fiph + 1);
while (rlen > 0) {
dlen = desc->fip_dlen * FIP_BPW;
if (dlen < sizeof(*desc) || dlen > rlen)
return -EINVAL;
/* Drop Adv if there are duplicate critical descriptors */
if ((desc->fip_dtype < 32) &&
!(desc_mask & 1U << desc->fip_dtype)) {
LIBFCOE_FIP_DBG(fip, "Duplicate Critical "
"Descriptors in FIP adv\n");
return -EINVAL;
}
switch (desc->fip_dtype) {
case FIP_DT_PRI:
if (dlen != sizeof(struct fip_pri_desc))
goto len_err;
fcf->pri = ((struct fip_pri_desc *)desc)->fd_pri;
desc_mask &= ~BIT(FIP_DT_PRI);
break;
case FIP_DT_MAC:
if (dlen != sizeof(struct fip_mac_desc))
goto len_err;
memcpy(fcf->fcf_mac,
((struct fip_mac_desc *)desc)->fd_mac,
ETH_ALEN);
memcpy(fcf->fcoe_mac, fcf->fcf_mac, ETH_ALEN);
if (!is_valid_ether_addr(fcf->fcf_mac)) {
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
LIBFCOE_FIP_DBG(fip,
"Invalid MAC addr %pM in FIP adv\n",
fcf->fcf_mac);
return -EINVAL;
}
desc_mask &= ~BIT(FIP_DT_MAC);
break;
case FIP_DT_NAME:
if (dlen != sizeof(struct fip_wwn_desc))
goto len_err;
wwn = (struct fip_wwn_desc *)desc;
fcf->switch_name = get_unaligned_be64(&wwn->fd_wwn);
desc_mask &= ~BIT(FIP_DT_NAME);
break;
case FIP_DT_FAB:
if (dlen != sizeof(struct fip_fab_desc))
goto len_err;
fab = (struct fip_fab_desc *)desc;
fcf->fabric_name = get_unaligned_be64(&fab->fd_wwn);
fcf->vfid = ntohs(fab->fd_vfid);
fcf->fc_map = ntoh24(fab->fd_map);
desc_mask &= ~BIT(FIP_DT_FAB);
break;
case FIP_DT_FKA:
if (dlen != sizeof(struct fip_fka_desc))
goto len_err;
fka = (struct fip_fka_desc *)desc;
if (fka->fd_flags & FIP_FKA_ADV_D)
fcf->fd_flags = 1;
t = ntohl(fka->fd_fka_period);
if (t >= FCOE_CTLR_MIN_FKA)
fcf->fka_period = msecs_to_jiffies(t);
desc_mask &= ~BIT(FIP_DT_FKA);
break;
case FIP_DT_MAP_OUI:
case FIP_DT_FCOE_SIZE:
case FIP_DT_FLOGI:
case FIP_DT_FDISC:
case FIP_DT_LOGO:
case FIP_DT_ELP:
default:
LIBFCOE_FIP_DBG(fip, "unexpected descriptor type %x "
"in FIP adv\n", desc->fip_dtype);
/* standard says ignore unknown descriptors >= 128 */
if (desc->fip_dtype < FIP_DT_NON_CRITICAL)
return -EINVAL;
break;
}
desc = (struct fip_desc *)((char *)desc + dlen);
rlen -= dlen;
}
if (!fcf->fc_map || (fcf->fc_map & 0x10000))
return -EINVAL;
if (!fcf->switch_name)
return -EINVAL;
if (desc_mask) {
LIBFCOE_FIP_DBG(fip, "adv missing descriptors mask %x\n",
desc_mask);
return -EINVAL;
}
return 0;
len_err:
LIBFCOE_FIP_DBG(fip, "FIP length error in descriptor type %x len %zu\n",
desc->fip_dtype, dlen);
return -EINVAL;
}
/**
* fcoe_ctlr_recv_adv() - Handle an incoming advertisement
* @fip: The FCoE controller receiving the advertisement
* @skb: The received FIP packet
*/
static void fcoe_ctlr_recv_adv(struct fcoe_ctlr *fip, struct sk_buff *skb)
{
struct fcoe_fcf *fcf;
struct fcoe_fcf new;
unsigned long sol_tov = msecs_to_jiffies(FCOE_CTRL_SOL_TOV);
int first = 0;
int mtu_valid;
int found = 0;
int rc = 0;
if (fcoe_ctlr_parse_adv(fip, skb, &new))
return;
mutex_lock(&fip->ctlr_mutex);
first = list_empty(&fip->fcfs);
list_for_each_entry(fcf, &fip->fcfs, list) {
if (fcf->switch_name == new.switch_name &&
fcf->fabric_name == new.fabric_name &&
fcf->fc_map == new.fc_map &&
ether_addr_equal(fcf->fcf_mac, new.fcf_mac)) {
found = 1;
break;
}
}
if (!found) {
if (fip->fcf_count >= FCOE_CTLR_FCF_LIMIT)
goto out;
fcf = kmalloc(sizeof(*fcf), GFP_ATOMIC);
if (!fcf)
goto out;
memcpy(fcf, &new, sizeof(new));
fcf->fip = fip;
rc = fcoe_sysfs_fcf_add(fcf);
if (rc) {
printk(KERN_ERR "Failed to allocate sysfs instance "
"for FCF, fab %16.16llx mac %pM\n",
new.fabric_name, new.fcf_mac);
kfree(fcf);
goto out;
}
} else {
/*
* Update the FCF's keep-alive descriptor flags.
* Other flag changes from new advertisements are
* ignored after a solicited advertisement is
* received and the FCF is selectable (usable).
*/
fcf->fd_flags = new.fd_flags;
if (!fcoe_ctlr_fcf_usable(fcf))
fcf->flags = new.flags;
if (fcf == fip->sel_fcf && !fcf->fd_flags) {
fip->ctlr_ka_time -= fcf->fka_period;
fip->ctlr_ka_time += new.fka_period;
if (time_before(fip->ctlr_ka_time, fip->timer.expires))
mod_timer(&fip->timer, fip->ctlr_ka_time);
}
fcf->fka_period = new.fka_period;
memcpy(fcf->fcf_mac, new.fcf_mac, ETH_ALEN);
}
mtu_valid = fcoe_ctlr_mtu_valid(fcf);
fcf->time = jiffies;
if (!found)
LIBFCOE_FIP_DBG(fip, "New FCF fab %16.16llx mac %pM\n",
fcf->fabric_name, fcf->fcf_mac);
/*
* If this advertisement is not solicited and our max receive size
* hasn't been verified, send a solicited advertisement.
*/
if (!mtu_valid)
fcoe_ctlr_solicit(fip, fcf);
/*
* If its been a while since we did a solicit, and this is
* the first advertisement we've received, do a multicast
* solicitation to gather as many advertisements as we can
* before selection occurs.
*/
if (first && time_after(jiffies, fip->sol_time + sol_tov))
fcoe_ctlr_solicit(fip, NULL);
/*
* Put this FCF at the head of the list for priority among equals.
* This helps in the case of an NPV switch which insists we use
* the FCF that answers multicast solicitations, not the others that
* are sending periodic multicast advertisements.
*/
if (mtu_valid)
list_move(&fcf->list, &fip->fcfs);
/*
* If this is the first validated FCF, note the time and
* set a timer to trigger selection.
*/
if (mtu_valid && !fip->sel_fcf && !fip->sel_time &&
fcoe_ctlr_fcf_usable(fcf)) {
fip->sel_time = jiffies +
msecs_to_jiffies(FCOE_CTLR_START_DELAY);
if (!timer_pending(&fip->timer) ||
time_before(fip->sel_time, fip->timer.expires))
mod_timer(&fip->timer, fip->sel_time);
}
out:
mutex_unlock(&fip->ctlr_mutex);
}
/**
* fcoe_ctlr_recv_els() - Handle an incoming FIP encapsulated ELS frame
* @fip: The FCoE controller which received the packet
* @skb: The received FIP packet
*/
static void fcoe_ctlr_recv_els(struct fcoe_ctlr *fip, struct sk_buff *skb)
{
struct fc_lport *lport = fip->lp;
struct fip_header *fiph;
struct fc_frame *fp = (struct fc_frame *)skb;
struct fc_frame_header *fh = NULL;
struct fip_desc *desc;
struct fip_encaps *els;
struct fcoe_fcf *sel;
struct fc_stats *stats;
enum fip_desc_type els_dtype = 0;
u8 els_op;
u8 sub;
u8 granted_mac[ETH_ALEN] = { 0 };
size_t els_len = 0;
size_t rlen;
size_t dlen;
u32 desc_mask = 0;
u32 desc_cnt = 0;
fiph = (struct fip_header *)skb->data;
sub = fiph->fip_subcode;
if (sub != FIP_SC_REQ && sub != FIP_SC_REP)
goto drop;
rlen = ntohs(fiph->fip_dl_len) * 4;
if (rlen + sizeof(*fiph) > skb->len)
goto drop;
desc = (struct fip_desc *)(fiph + 1);
while (rlen > 0) {
desc_cnt++;
dlen = desc->fip_dlen * FIP_BPW;
if (dlen < sizeof(*desc) || dlen > rlen)
goto drop;
/* Drop ELS if there are duplicate critical descriptors */
if (desc->fip_dtype < 32) {
if ((desc->fip_dtype != FIP_DT_MAC) &&
(desc_mask & 1U << desc->fip_dtype)) {
LIBFCOE_FIP_DBG(fip, "Duplicate Critical "
"Descriptors in FIP ELS\n");
goto drop;
}
desc_mask |= (1 << desc->fip_dtype);
}
switch (desc->fip_dtype) {
case FIP_DT_MAC:
sel = fip->sel_fcf;
if (desc_cnt == 1) {
LIBFCOE_FIP_DBG(fip, "FIP descriptors "
"received out of order\n");
goto drop;
}
/*
* Some switch implementations send two MAC descriptors,
* with first MAC(granted_mac) being the FPMA, and the
* second one(fcoe_mac) is used as destination address
* for sending/receiving FCoE packets. FIP traffic is
* sent using fip_mac. For regular switches, both
* fip_mac and fcoe_mac would be the same.
*/
if (desc_cnt == 2)
memcpy(granted_mac,
((struct fip_mac_desc *)desc)->fd_mac,
ETH_ALEN);
if (dlen != sizeof(struct fip_mac_desc))
goto len_err;
if ((desc_cnt == 3) && (sel))
memcpy(sel->fcoe_mac,
((struct fip_mac_desc *)desc)->fd_mac,
ETH_ALEN);
break;
case FIP_DT_FLOGI:
case FIP_DT_FDISC:
case FIP_DT_LOGO:
case FIP_DT_ELP:
if (desc_cnt != 1) {
LIBFCOE_FIP_DBG(fip, "FIP descriptors "
"received out of order\n");
goto drop;
}
if (fh)
goto drop;
if (dlen < sizeof(*els) + sizeof(*fh) + 1)
goto len_err;
els_len = dlen - sizeof(*els);
els = (struct fip_encaps *)desc;
fh = (struct fc_frame_header *)(els + 1);
els_dtype = desc->fip_dtype;
break;
default:
LIBFCOE_FIP_DBG(fip, "unexpected descriptor type %x "
"in FIP adv\n", desc->fip_dtype);
/* standard says ignore unknown descriptors >= 128 */
if (desc->fip_dtype < FIP_DT_NON_CRITICAL)
goto drop;
if (desc_cnt <= 2) {
LIBFCOE_FIP_DBG(fip, "FIP descriptors "
"received out of order\n");
goto drop;
}
break;
}
desc = (struct fip_desc *)((char *)desc + dlen);
rlen -= dlen;
}
if (!fh)
goto drop;
els_op = *(u8 *)(fh + 1);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if ((els_dtype == FIP_DT_FLOGI || els_dtype == FIP_DT_FDISC) &&
sub == FIP_SC_REP && fip->mode != FIP_MODE_VN2VN) {
if (els_op == ELS_LS_ACC) {
if (!is_valid_ether_addr(granted_mac)) {
LIBFCOE_FIP_DBG(fip,
"Invalid MAC address %pM in FIP ELS\n",
granted_mac);
goto drop;
}
memcpy(fr_cb(fp)->granted_mac, granted_mac, ETH_ALEN);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (fip->flogi_oxid == ntohs(fh->fh_ox_id)) {
fip->flogi_oxid = FC_XID_UNKNOWN;
if (els_dtype == FIP_DT_FLOGI)
fcoe_ctlr_announce(fip);
}
} else if (els_dtype == FIP_DT_FLOGI &&
!fcoe_ctlr_flogi_retry(fip))
goto drop; /* retrying FLOGI so drop reject */
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
}
if ((desc_cnt == 0) || ((els_op != ELS_LS_RJT) &&
(!(1U << FIP_DT_MAC & desc_mask)))) {
LIBFCOE_FIP_DBG(fip, "Missing critical descriptors "
"in FIP ELS\n");
goto drop;
}
/*
* Convert skb into an fc_frame containing only the ELS.
*/
skb_pull(skb, (u8 *)fh - skb->data);
skb_trim(skb, els_len);
fp = (struct fc_frame *)skb;
fc_frame_init(fp);
fr_sof(fp) = FC_SOF_I3;
fr_eof(fp) = FC_EOF_T;
fr_dev(fp) = lport;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fr_encaps(fp) = els_dtype;
stats = per_cpu_ptr(lport->stats, get_cpu());
stats->RxFrames++;
stats->RxWords += skb->len / FIP_BPW;
put_cpu();
fc_exch_recv(lport, fp);
return;
len_err:
LIBFCOE_FIP_DBG(fip, "FIP length error in descriptor type %x len %zu\n",
desc->fip_dtype, dlen);
drop:
kfree_skb(skb);
}
/**
* fcoe_ctlr_recv_els() - Handle an incoming link reset frame
* @fip: The FCoE controller that received the frame
* @fh: The received FIP header
*
* There may be multiple VN_Port descriptors.
* The overall length has already been checked.
*/
static void fcoe_ctlr_recv_clr_vlink(struct fcoe_ctlr *fip,
struct sk_buff *skb)
{
struct fip_desc *desc;
struct fip_mac_desc *mp;
struct fip_wwn_desc *wp;
struct fip_vn_desc *vp;
size_t rlen;
size_t dlen;
struct fcoe_fcf *fcf = fip->sel_fcf;
struct fc_lport *lport = fip->lp;
struct fc_lport *vn_port = NULL;
u32 desc_mask;
int num_vlink_desc;
int reset_phys_port = 0;
struct fip_vn_desc **vlink_desc_arr = NULL;
struct fip_header *fh = (struct fip_header *)skb->data;
struct ethhdr *eh = eth_hdr(skb);
LIBFCOE_FIP_DBG(fip, "Clear Virtual Link received\n");
if (!fcf) {
/*
* We are yet to select best FCF, but we got CVL in the
* meantime. reset the ctlr and let it rediscover the FCF
*/
LIBFCOE_FIP_DBG(fip, "Resetting fcoe_ctlr as FCF has not been "
"selected yet\n");
mutex_lock(&fip->ctlr_mutex);
fcoe_ctlr_reset(fip);
mutex_unlock(&fip->ctlr_mutex);
return;
}
/*
* If we've selected an FCF check that the CVL is from there to avoid
* processing CVLs from an unexpected source. If it is from an
* unexpected source drop it on the floor.
*/
if (!ether_addr_equal(eh->h_source, fcf->fcf_mac)) {
LIBFCOE_FIP_DBG(fip, "Dropping CVL due to source address "
"mismatch with FCF src=%pM\n", eh->h_source);
return;
}
/*
* If we haven't logged into the fabric but receive a CVL we should
* reset everything and go back to solicitation.
*/
if (!lport->port_id) {
LIBFCOE_FIP_DBG(fip, "lport not logged in, resoliciting\n");
mutex_lock(&fip->ctlr_mutex);
fcoe_ctlr_reset(fip);
mutex_unlock(&fip->ctlr_mutex);
fc_lport_reset(fip->lp);
fcoe_ctlr_solicit(fip, NULL);
return;
}
/*
* mask of required descriptors. Validating each one clears its bit.
*/
desc_mask = BIT(FIP_DT_MAC) | BIT(FIP_DT_NAME);
rlen = ntohs(fh->fip_dl_len) * FIP_BPW;
desc = (struct fip_desc *)(fh + 1);
/*
* Actually need to subtract 'sizeof(*mp) - sizeof(*wp)' from 'rlen'
* before determining max Vx_Port descriptor but a buggy FCF could have
* omitted either or both MAC Address and Name Identifier descriptors
*/
num_vlink_desc = rlen / sizeof(*vp);
if (num_vlink_desc)
treewide: kmalloc() -> kmalloc_array() The kmalloc() function has a 2-factor argument form, kmalloc_array(). This patch replaces cases of: kmalloc(a * b, gfp) with: kmalloc_array(a * b, gfp) as well as handling cases of: kmalloc(a * b * c, gfp) with: kmalloc(array3_size(a, b, c), gfp) as it's slightly less ugly than: kmalloc_array(array_size(a, b), c, gfp) This does, however, attempt to ignore constant size factors like: kmalloc(4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The tools/ directory was manually excluded, since it has its own implementation of kmalloc(). The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ type TYPE; expression THING, E; @@ ( kmalloc( - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | kmalloc( - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression COUNT; typedef u8; typedef __u8; @@ ( kmalloc( - sizeof(u8) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(__u8) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(char) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(unsigned char) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(u8) * COUNT + COUNT , ...) | kmalloc( - sizeof(__u8) * COUNT + COUNT , ...) | kmalloc( - sizeof(char) * COUNT + COUNT , ...) | kmalloc( - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( - kmalloc + kmalloc_array ( - sizeof(TYPE) * (COUNT_ID) + COUNT_ID, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * COUNT_ID + COUNT_ID, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * (COUNT_CONST) + COUNT_CONST, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * COUNT_CONST + COUNT_CONST, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * (COUNT_ID) + COUNT_ID, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * COUNT_ID + COUNT_ID, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * (COUNT_CONST) + COUNT_CONST, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * COUNT_CONST + COUNT_CONST, sizeof(THING) , ...) ) // 2-factor product, only identifiers. @@ identifier SIZE, COUNT; @@ - kmalloc + kmalloc_array ( - SIZE * COUNT + COUNT, SIZE , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( kmalloc( - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kmalloc( - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kmalloc( - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kmalloc( - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( kmalloc( - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kmalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kmalloc( - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kmalloc( - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kmalloc( - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | kmalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ identifier STRIDE, SIZE, COUNT; @@ ( kmalloc( - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products, // when they're not all constants... @@ expression E1, E2, E3; constant C1, C2, C3; @@ ( kmalloc(C1 * C2 * C3, ...) | kmalloc( - (E1) * E2 * E3 + array3_size(E1, E2, E3) , ...) | kmalloc( - (E1) * (E2) * E3 + array3_size(E1, E2, E3) , ...) | kmalloc( - (E1) * (E2) * (E3) + array3_size(E1, E2, E3) , ...) | kmalloc( - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants, // keeping sizeof() as the second factor argument. @@ expression THING, E1, E2; type TYPE; constant C1, C2, C3; @@ ( kmalloc(sizeof(THING) * C2, ...) | kmalloc(sizeof(TYPE) * C2, ...) | kmalloc(C1 * C2 * C3, ...) | kmalloc(C1 * C2, ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * (E2) + E2, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * E2 + E2, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * (E2) + E2, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * E2 + E2, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - (E1) * E2 + E1, E2 , ...) | - kmalloc + kmalloc_array ( - (E1) * (E2) + E1, E2 , ...) | - kmalloc + kmalloc_array ( - E1 * E2 + E1, E2 , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-13 04:55:00 +08:00
vlink_desc_arr = kmalloc_array(num_vlink_desc, sizeof(vp),
GFP_ATOMIC);
if (!vlink_desc_arr)
return;
num_vlink_desc = 0;
while (rlen >= sizeof(*desc)) {
dlen = desc->fip_dlen * FIP_BPW;
if (dlen > rlen)
goto err;
/* Drop CVL if there are duplicate critical descriptors */
if ((desc->fip_dtype < 32) &&
(desc->fip_dtype != FIP_DT_VN_ID) &&
!(desc_mask & 1U << desc->fip_dtype)) {
LIBFCOE_FIP_DBG(fip, "Duplicate Critical "
"Descriptors in FIP CVL\n");
goto err;
}
switch (desc->fip_dtype) {
case FIP_DT_MAC:
mp = (struct fip_mac_desc *)desc;
if (dlen < sizeof(*mp))
goto err;
if (!ether_addr_equal(mp->fd_mac, fcf->fcf_mac))
goto err;
desc_mask &= ~BIT(FIP_DT_MAC);
break;
case FIP_DT_NAME:
wp = (struct fip_wwn_desc *)desc;
if (dlen < sizeof(*wp))
goto err;
if (get_unaligned_be64(&wp->fd_wwn) != fcf->switch_name)
goto err;
desc_mask &= ~BIT(FIP_DT_NAME);
break;
case FIP_DT_VN_ID:
vp = (struct fip_vn_desc *)desc;
if (dlen < sizeof(*vp))
goto err;
vlink_desc_arr[num_vlink_desc++] = vp;
vn_port = fc_vport_id_lookup(lport,
ntoh24(vp->fd_fc_id));
if (vn_port && (vn_port == lport)) {
mutex_lock(&fip->ctlr_mutex);
per_cpu_ptr(lport->stats,
get_cpu())->VLinkFailureCount++;
put_cpu();
fcoe_ctlr_reset(fip);
mutex_unlock(&fip->ctlr_mutex);
}
break;
default:
/* standard says ignore unknown descriptors >= 128 */
if (desc->fip_dtype < FIP_DT_NON_CRITICAL)
goto err;
break;
}
desc = (struct fip_desc *)((char *)desc + dlen);
rlen -= dlen;
}
/*
* reset only if all required descriptors were present and valid.
*/
if (desc_mask)
LIBFCOE_FIP_DBG(fip, "missing descriptors mask %x\n",
desc_mask);
else if (!num_vlink_desc) {
LIBFCOE_FIP_DBG(fip, "CVL: no Vx_Port descriptor found\n");
/*
* No Vx_Port description. Clear all NPIV ports,
* followed by physical port
*/
mutex_lock(&fip->ctlr_mutex);
per_cpu_ptr(lport->stats, get_cpu())->VLinkFailureCount++;
put_cpu();
fcoe_ctlr_reset(fip);
mutex_unlock(&fip->ctlr_mutex);
mutex_lock(&lport->lp_mutex);
list_for_each_entry(vn_port, &lport->vports, list)
fc_lport_reset(vn_port);
mutex_unlock(&lport->lp_mutex);
fc_lport_reset(fip->lp);
fcoe_ctlr_solicit(fip, NULL);
} else {
int i;
LIBFCOE_FIP_DBG(fip, "performing Clear Virtual Link\n");
for (i = 0; i < num_vlink_desc; i++) {
vp = vlink_desc_arr[i];
vn_port = fc_vport_id_lookup(lport,
ntoh24(vp->fd_fc_id));
if (!vn_port)
continue;
/*
* 'port_id' is already validated, check MAC address and
* wwpn
*/
if (!ether_addr_equal(fip->get_src_addr(vn_port),
vp->fd_mac) ||
get_unaligned_be64(&vp->fd_wwpn) !=
vn_port->wwpn)
continue;
if (vn_port == lport)
/*
* Physical port, defer processing till all
* listed NPIV ports are cleared
*/
reset_phys_port = 1;
else /* NPIV port */
fc_lport_reset(vn_port);
}
if (reset_phys_port) {
fc_lport_reset(fip->lp);
fcoe_ctlr_solicit(fip, NULL);
}
}
err:
kfree(vlink_desc_arr);
}
/**
* fcoe_ctlr_recv() - Receive a FIP packet
* @fip: The FCoE controller that received the packet
* @skb: The received FIP packet
*
* This may be called from either NET_RX_SOFTIRQ or IRQ.
*/
void fcoe_ctlr_recv(struct fcoe_ctlr *fip, struct sk_buff *skb)
{
fcoe: ensure that skb placed on the fip_recv_list are unshared Recently had this Oops reported to me on the 3.10 kernel: [ 807.554955] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 [ 807.562799] IP: [<ffffffff814e6fc7>] skb_dequeue+0x47/0x70 [ 807.568296] PGD 20c889067 PUD 20c8b8067 PMD 0 [ 807.572769] Oops: 0002 [#1] SMP [ 807.655597] Hardware name: Dell Inc. PowerEdge R415/0DDT2D, BIOS 1.8.6 12/06/2011 [ 807.663079] Workqueue: events fcoe_ctlr_recv_work [libfcoe] [ 807.668656] task: ffff88020b42a160 ti: ffff88020ae6c000 task.ti: ffff88020ae6c000 [ 807.676126] RIP: 0010:[<ffffffff814e6fc7>] [<ffffffff814e6fc7>] skb_dequeue+0x47/0x70 [ 807.684046] RSP: 0000:ffff88020ae6dd70 EFLAGS: 00010097 [ 807.689349] RAX: 0000000000000246 RBX: ffff8801d04d6700 RCX: 0000000000000000 [ 807.696474] RDX: 0000000000000000 RSI: 0000000000000246 RDI: ffff88020df26434 [ 807.703598] RBP: ffff88020ae6dd88 R08: 00000000000173e0 R09: ffff880216e173e0 [ 807.710723] R10: ffffffff814e5897 R11: ffffea0007413580 R12: ffff88020df26420 [ 807.717847] R13: ffff88020df26434 R14: 0000000000000004 R15: ffff8801d04c42ce [ 807.724972] FS: 00007fdaab6048c0(0000) GS:ffff880216e00000(0000) knlGS:0000000000000000 [ 807.733049] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 807.738785] CR2: 0000000000000008 CR3: 000000020cbc9000 CR4: 00000000000006f0 [ 807.745910] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 807.753033] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 807.760156] Stack: [ 807.762162] ffff8801d04d6700 0000000000000001 ffff88020df26400 ffff88020ae6de20 [ 807.769586] ffffffffa0444409 ffff88020b046a00 ffff88020ae6dde8 ffffffff810105be [ 807.777008] ffff88020b42a868 0000000000000000 ffff88020df264a8 ffff88020df26348 [ 807.784431] Call Trace: [ 807.786885] [<ffffffffa0444409>] fcoe_ctlr_recv_work+0x59/0x9a0 [libfcoe] [ 807.793755] [<ffffffff810105be>] ? __switch_to+0x13e/0x4a0 [ 807.799324] [<ffffffff8107d0e6>] process_one_work+0x176/0x420 [ 807.805151] [<ffffffff8107dd0b>] worker_thread+0x11b/0x3a0 [ 807.810717] [<ffffffff8107dbf0>] ? rescuer_thread+0x350/0x350 [ 807.816545] [<ffffffff810842b0>] kthread+0xc0/0xd0 [ 807.821416] [<ffffffff810841f0>] ? insert_kthread_work+0x40/0x40 [ 807.827503] [<ffffffff8160ce2c>] ret_from_fork+0x7c/0xb0 [ 807.832897] [<ffffffff810841f0>] ? insert_kthread_work+0x40/0x40 [ 807.858500] RIP [<ffffffff814e6fc7>] skb_dequeue+0x47/0x70 [ 807.864076] RSP <ffff88020ae6dd70> [ 807.867558] CR2: 0000000000000008 Looks like the root cause is the fact that the packet recieve function fcoe_ctlr_recv enqueues the skb to a sk_buff_head_list prior to ensuring that the skb is unshared. This can happen when multiple packet listeners recieve an skb, as the deliver_skb function just increments skb->users for each handler. As a result, having multiple users of a single skb results in multiple manipulators of its methods, implying list corruption, and the oops recorded above. The fix is pretty easy, just make sure that we clone the skb if its got multiple users with the skb_share_check function, like other protocols do. Signed-off-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: Robert Love <robert.w.love@intel.com>
2013-08-03 18:45:53 +08:00
skb = skb_share_check(skb, GFP_ATOMIC);
if (!skb)
return;
skb_queue_tail(&fip->fip_recv_list, skb);
schedule_work(&fip->recv_work);
}
EXPORT_SYMBOL(fcoe_ctlr_recv);
/**
* fcoe_ctlr_recv_handler() - Receive a FIP frame
* @fip: The FCoE controller that received the frame
* @skb: The received FIP frame
*
* Returns non-zero if the frame is dropped.
*/
static int fcoe_ctlr_recv_handler(struct fcoe_ctlr *fip, struct sk_buff *skb)
{
struct fip_header *fiph;
struct ethhdr *eh;
enum fip_state state;
bool fip_vlan_resp = false;
u16 op;
u8 sub;
if (skb_linearize(skb))
goto drop;
if (skb->len < sizeof(*fiph))
goto drop;
eh = eth_hdr(skb);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (fip->mode == FIP_MODE_VN2VN) {
if (!ether_addr_equal(eh->h_dest, fip->ctl_src_addr) &&
!ether_addr_equal(eh->h_dest, fcoe_all_vn2vn) &&
!ether_addr_equal(eh->h_dest, fcoe_all_p2p))
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
goto drop;
} else if (!ether_addr_equal(eh->h_dest, fip->ctl_src_addr) &&
!ether_addr_equal(eh->h_dest, fcoe_all_enode))
goto drop;
fiph = (struct fip_header *)skb->data;
op = ntohs(fiph->fip_op);
sub = fiph->fip_subcode;
if (FIP_VER_DECAPS(fiph->fip_ver) != FIP_VER)
goto drop;
if (ntohs(fiph->fip_dl_len) * FIP_BPW + sizeof(*fiph) > skb->len)
goto drop;
mutex_lock(&fip->ctlr_mutex);
state = fip->state;
if (state == FIP_ST_AUTO) {
fip->map_dest = 0;
fcoe_ctlr_set_state(fip, FIP_ST_ENABLED);
state = FIP_ST_ENABLED;
LIBFCOE_FIP_DBG(fip, "Using FIP mode\n");
}
fip_vlan_resp = fip->fip_resp;
mutex_unlock(&fip->ctlr_mutex);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (fip->mode == FIP_MODE_VN2VN && op == FIP_OP_VN2VN)
return fcoe_ctlr_vn_recv(fip, skb);
if (fip_vlan_resp && op == FIP_OP_VLAN) {
LIBFCOE_FIP_DBG(fip, "fip vlan discovery\n");
return fcoe_ctlr_vlan_recv(fip, skb);
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (state != FIP_ST_ENABLED && state != FIP_ST_VNMP_UP &&
state != FIP_ST_VNMP_CLAIM)
goto drop;
if (op == FIP_OP_LS) {
fcoe_ctlr_recv_els(fip, skb); /* consumes skb */
return 0;
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (state != FIP_ST_ENABLED)
goto drop;
if (op == FIP_OP_DISC && sub == FIP_SC_ADV)
fcoe_ctlr_recv_adv(fip, skb);
else if (op == FIP_OP_CTRL && sub == FIP_SC_CLR_VLINK)
fcoe_ctlr_recv_clr_vlink(fip, skb);
kfree_skb(skb);
return 0;
drop:
kfree_skb(skb);
return -1;
}
/**
* fcoe_ctlr_select() - Select the best FCF (if possible)
* @fip: The FCoE controller
*
* Returns the selected FCF, or NULL if none are usable.
*
* If there are conflicting advertisements, no FCF can be chosen.
*
* If there is already a selected FCF, this will choose a better one or
* an equivalent one that hasn't already been sent a FLOGI.
*
* Called with lock held.
*/
static struct fcoe_fcf *fcoe_ctlr_select(struct fcoe_ctlr *fip)
{
struct fcoe_fcf *fcf;
struct fcoe_fcf *best = fip->sel_fcf;
list_for_each_entry(fcf, &fip->fcfs, list) {
LIBFCOE_FIP_DBG(fip, "consider FCF fab %16.16llx "
"VFID %d mac %pM map %x val %d "
"sent %u pri %u\n",
fcf->fabric_name, fcf->vfid, fcf->fcf_mac,
fcf->fc_map, fcoe_ctlr_mtu_valid(fcf),
fcf->flogi_sent, fcf->pri);
if (!fcoe_ctlr_fcf_usable(fcf)) {
LIBFCOE_FIP_DBG(fip, "FCF for fab %16.16llx "
"map %x %svalid %savailable\n",
fcf->fabric_name, fcf->fc_map,
(fcf->flags & FIP_FL_SOL) ? "" : "in",
(fcf->flags & FIP_FL_AVAIL) ?
"" : "un");
continue;
}
if (!best || fcf->pri < best->pri || best->flogi_sent)
best = fcf;
if (fcf->fabric_name != best->fabric_name ||
fcf->vfid != best->vfid ||
fcf->fc_map != best->fc_map) {
LIBFCOE_FIP_DBG(fip, "Conflicting fabric, VFID, "
"or FC-MAP\n");
return NULL;
}
}
fip->sel_fcf = best;
if (best) {
LIBFCOE_FIP_DBG(fip, "using FCF mac %pM\n", best->fcf_mac);
fip->port_ka_time = jiffies +
msecs_to_jiffies(FIP_VN_KA_PERIOD);
fip->ctlr_ka_time = jiffies + best->fka_period;
if (time_before(fip->ctlr_ka_time, fip->timer.expires))
mod_timer(&fip->timer, fip->ctlr_ka_time);
}
return best;
}
/**
* fcoe_ctlr_flogi_send_locked() - send FIP-encapsulated FLOGI to current FCF
* @fip: The FCoE controller
*
* Returns non-zero error if it could not be sent.
*
* Called with ctlr_mutex and ctlr_lock held.
* Caller must verify that fip->sel_fcf is not NULL.
*/
static int fcoe_ctlr_flogi_send_locked(struct fcoe_ctlr *fip)
{
struct sk_buff *skb;
struct sk_buff *skb_orig;
struct fc_frame_header *fh;
int error;
skb_orig = fip->flogi_req;
if (!skb_orig)
return -EINVAL;
/*
* Clone and send the FLOGI request. If clone fails, use original.
*/
skb = skb_clone(skb_orig, GFP_ATOMIC);
if (!skb) {
skb = skb_orig;
fip->flogi_req = NULL;
}
fh = (struct fc_frame_header *)skb->data;
error = fcoe_ctlr_encaps(fip, fip->lp, FIP_DT_FLOGI, skb,
ntoh24(fh->fh_d_id));
if (error) {
kfree_skb(skb);
return error;
}
fip->send(fip, skb);
fip->sel_fcf->flogi_sent = 1;
return 0;
}
/**
* fcoe_ctlr_flogi_retry() - resend FLOGI request to a new FCF if possible
* @fip: The FCoE controller
*
* Returns non-zero error code if there's no FLOGI request to retry or
* no alternate FCF available.
*/
static int fcoe_ctlr_flogi_retry(struct fcoe_ctlr *fip)
{
struct fcoe_fcf *fcf;
int error;
mutex_lock(&fip->ctlr_mutex);
spin_lock_bh(&fip->ctlr_lock);
LIBFCOE_FIP_DBG(fip, "re-sending FLOGI - reselect\n");
fcf = fcoe_ctlr_select(fip);
if (!fcf || fcf->flogi_sent) {
kfree_skb(fip->flogi_req);
fip->flogi_req = NULL;
error = -ENOENT;
} else {
fcoe_ctlr_solicit(fip, NULL);
error = fcoe_ctlr_flogi_send_locked(fip);
}
spin_unlock_bh(&fip->ctlr_lock);
mutex_unlock(&fip->ctlr_mutex);
return error;
}
/**
* fcoe_ctlr_flogi_send() - Handle sending of FIP FLOGI.
* @fip: The FCoE controller that timed out
*
* Done here because fcoe_ctlr_els_send() can't get mutex.
*
* Called with ctlr_mutex held. The caller must not hold ctlr_lock.
*/
static void fcoe_ctlr_flogi_send(struct fcoe_ctlr *fip)
{
struct fcoe_fcf *fcf;
spin_lock_bh(&fip->ctlr_lock);
fcf = fip->sel_fcf;
if (!fcf || !fip->flogi_req_send)
goto unlock;
LIBFCOE_FIP_DBG(fip, "sending FLOGI\n");
/*
* If this FLOGI is being sent due to a timeout retry
* to the same FCF as before, select a different FCF if possible.
*/
if (fcf->flogi_sent) {
LIBFCOE_FIP_DBG(fip, "sending FLOGI - reselect\n");
fcf = fcoe_ctlr_select(fip);
if (!fcf || fcf->flogi_sent) {
LIBFCOE_FIP_DBG(fip, "sending FLOGI - clearing\n");
list_for_each_entry(fcf, &fip->fcfs, list)
fcf->flogi_sent = 0;
fcf = fcoe_ctlr_select(fip);
}
}
if (fcf) {
fcoe_ctlr_flogi_send_locked(fip);
fip->flogi_req_send = 0;
} else /* XXX */
LIBFCOE_FIP_DBG(fip, "No FCF selected - defer send\n");
unlock:
spin_unlock_bh(&fip->ctlr_lock);
}
/**
* fcoe_ctlr_timeout() - FIP timeout handler
* @arg: The FCoE controller that timed out
*/
treewide: setup_timer() -> timer_setup() This converts all remaining cases of the old setup_timer() API into using timer_setup(), where the callback argument is the structure already holding the struct timer_list. These should have no behavioral changes, since they just change which pointer is passed into the callback with the same available pointers after conversion. It handles the following examples, in addition to some other variations. Casting from unsigned long: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... setup_timer(&ptr->my_timer, my_callback, ptr); and forced object casts: void my_callback(struct something *ptr) { ... } ... setup_timer(&ptr->my_timer, my_callback, (unsigned long)ptr); become: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... timer_setup(&ptr->my_timer, my_callback, 0); Direct function assignments: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... ptr->my_timer.function = my_callback; have a temporary cast added, along with converting the args: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... ptr->my_timer.function = (TIMER_FUNC_TYPE)my_callback; And finally, callbacks without a data assignment: void my_callback(unsigned long data) { ... } ... setup_timer(&ptr->my_timer, my_callback, 0); have their argument renamed to verify they're unused during conversion: void my_callback(struct timer_list *unused) { ... } ... timer_setup(&ptr->my_timer, my_callback, 0); The conversion is done with the following Coccinelle script: spatch --very-quiet --all-includes --include-headers \ -I ./arch/x86/include -I ./arch/x86/include/generated \ -I ./include -I ./arch/x86/include/uapi \ -I ./arch/x86/include/generated/uapi -I ./include/uapi \ -I ./include/generated/uapi --include ./include/linux/kconfig.h \ --dir . \ --cocci-file ~/src/data/timer_setup.cocci @fix_address_of@ expression e; @@ setup_timer( -&(e) +&e , ...) // Update any raw setup_timer() usages that have a NULL callback, but // would otherwise match change_timer_function_usage, since the latter // will update all function assignments done in the face of a NULL // function initialization in setup_timer(). @change_timer_function_usage_NULL@ expression _E; identifier _timer; type _cast_data; @@ ( -setup_timer(&_E->_timer, NULL, _E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E->_timer, NULL, (_cast_data)_E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E._timer, NULL, &_E); +timer_setup(&_E._timer, NULL, 0); | -setup_timer(&_E._timer, NULL, (_cast_data)&_E); +timer_setup(&_E._timer, NULL, 0); ) @change_timer_function_usage@ expression _E; identifier _timer; struct timer_list _stl; identifier _callback; type _cast_func, _cast_data; @@ ( -setup_timer(&_E->_timer, _callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | _E->_timer@_stl.function = _callback; | _E->_timer@_stl.function = &_callback; | _E->_timer@_stl.function = (_cast_func)_callback; | _E->_timer@_stl.function = (_cast_func)&_callback; | _E._timer@_stl.function = _callback; | _E._timer@_stl.function = &_callback; | _E._timer@_stl.function = (_cast_func)_callback; | _E._timer@_stl.function = (_cast_func)&_callback; ) // callback(unsigned long arg) @change_callback_handle_cast depends on change_timer_function_usage@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; identifier _handle; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { ( ... when != _origarg _handletype *_handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg ) } // callback(unsigned long arg) without existing variable @change_callback_handle_cast_no_arg depends on change_timer_function_usage && !change_callback_handle_cast@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { + _handletype *_origarg = from_timer(_origarg, t, _timer); + ... when != _origarg - (_handletype *)_origarg + _origarg ... when != _origarg } // Avoid already converted callbacks. @match_callback_converted depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier t; @@ void _callback(struct timer_list *t) { ... } // callback(struct something *handle) @change_callback_handle_arg depends on change_timer_function_usage && !match_callback_converted && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; @@ void _callback( -_handletype *_handle +struct timer_list *t ) { + _handletype *_handle = from_timer(_handle, t, _timer); ... } // If change_callback_handle_arg ran on an empty function, remove // the added handler. @unchange_callback_handle_arg depends on change_timer_function_usage && change_callback_handle_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; identifier t; @@ void _callback(struct timer_list *t) { - _handletype *_handle = from_timer(_handle, t, _timer); } // We only want to refactor the setup_timer() data argument if we've found // the matching callback. This undoes changes in change_timer_function_usage. @unchange_timer_function_usage depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg && !change_callback_handle_arg@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type change_timer_function_usage._cast_data; @@ ( -timer_setup(&_E->_timer, _callback, 0); +setup_timer(&_E->_timer, _callback, (_cast_data)_E); | -timer_setup(&_E._timer, _callback, 0); +setup_timer(&_E._timer, _callback, (_cast_data)&_E); ) // If we fixed a callback from a .function assignment, fix the // assignment cast now. @change_timer_function_assignment depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_func; typedef TIMER_FUNC_TYPE; @@ ( _E->_timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -&_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)_callback; +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -&_callback; +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; ) // Sometimes timer functions are called directly. Replace matched args. @change_timer_function_calls depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression _E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_data; @@ _callback( ( -(_cast_data)_E +&_E->_timer | -(_cast_data)&_E +&_E._timer | -_E +&_E->_timer ) ) // If a timer has been configured without a data argument, it can be // converted without regard to the callback argument, since it is unused. @match_timer_function_unused_data@ expression _E; identifier _timer; identifier _callback; @@ ( -setup_timer(&_E->_timer, _callback, 0); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0L); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0UL); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0L); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0UL); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_timer, _callback, 0); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0L); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0UL); +timer_setup(&_timer, _callback, 0); | -setup_timer(_timer, _callback, 0); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0L); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0UL); +timer_setup(_timer, _callback, 0); ) @change_callback_unused_data depends on match_timer_function_unused_data@ identifier match_timer_function_unused_data._callback; type _origtype; identifier _origarg; @@ void _callback( -_origtype _origarg +struct timer_list *unused ) { ... when != _origarg } Signed-off-by: Kees Cook <keescook@chromium.org>
2017-10-17 05:43:17 +08:00
static void fcoe_ctlr_timeout(struct timer_list *t)
{
treewide: setup_timer() -> timer_setup() This converts all remaining cases of the old setup_timer() API into using timer_setup(), where the callback argument is the structure already holding the struct timer_list. These should have no behavioral changes, since they just change which pointer is passed into the callback with the same available pointers after conversion. It handles the following examples, in addition to some other variations. Casting from unsigned long: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... setup_timer(&ptr->my_timer, my_callback, ptr); and forced object casts: void my_callback(struct something *ptr) { ... } ... setup_timer(&ptr->my_timer, my_callback, (unsigned long)ptr); become: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... timer_setup(&ptr->my_timer, my_callback, 0); Direct function assignments: void my_callback(unsigned long data) { struct something *ptr = (struct something *)data; ... } ... ptr->my_timer.function = my_callback; have a temporary cast added, along with converting the args: void my_callback(struct timer_list *t) { struct something *ptr = from_timer(ptr, t, my_timer); ... } ... ptr->my_timer.function = (TIMER_FUNC_TYPE)my_callback; And finally, callbacks without a data assignment: void my_callback(unsigned long data) { ... } ... setup_timer(&ptr->my_timer, my_callback, 0); have their argument renamed to verify they're unused during conversion: void my_callback(struct timer_list *unused) { ... } ... timer_setup(&ptr->my_timer, my_callback, 0); The conversion is done with the following Coccinelle script: spatch --very-quiet --all-includes --include-headers \ -I ./arch/x86/include -I ./arch/x86/include/generated \ -I ./include -I ./arch/x86/include/uapi \ -I ./arch/x86/include/generated/uapi -I ./include/uapi \ -I ./include/generated/uapi --include ./include/linux/kconfig.h \ --dir . \ --cocci-file ~/src/data/timer_setup.cocci @fix_address_of@ expression e; @@ setup_timer( -&(e) +&e , ...) // Update any raw setup_timer() usages that have a NULL callback, but // would otherwise match change_timer_function_usage, since the latter // will update all function assignments done in the face of a NULL // function initialization in setup_timer(). @change_timer_function_usage_NULL@ expression _E; identifier _timer; type _cast_data; @@ ( -setup_timer(&_E->_timer, NULL, _E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E->_timer, NULL, (_cast_data)_E); +timer_setup(&_E->_timer, NULL, 0); | -setup_timer(&_E._timer, NULL, &_E); +timer_setup(&_E._timer, NULL, 0); | -setup_timer(&_E._timer, NULL, (_cast_data)&_E); +timer_setup(&_E._timer, NULL, 0); ) @change_timer_function_usage@ expression _E; identifier _timer; struct timer_list _stl; identifier _callback; type _cast_func, _cast_data; @@ ( -setup_timer(&_E->_timer, _callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, &_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, _E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, &_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)_E); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, (_cast_func)&_callback, (_cast_data)&_E); +timer_setup(&_E._timer, _callback, 0); | _E->_timer@_stl.function = _callback; | _E->_timer@_stl.function = &_callback; | _E->_timer@_stl.function = (_cast_func)_callback; | _E->_timer@_stl.function = (_cast_func)&_callback; | _E._timer@_stl.function = _callback; | _E._timer@_stl.function = &_callback; | _E._timer@_stl.function = (_cast_func)_callback; | _E._timer@_stl.function = (_cast_func)&_callback; ) // callback(unsigned long arg) @change_callback_handle_cast depends on change_timer_function_usage@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; identifier _handle; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { ( ... when != _origarg _handletype *_handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(_handletype *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg | ... when != _origarg _handletype *_handle; ... when != _handle _handle = -(void *)_origarg; +from_timer(_handle, t, _timer); ... when != _origarg ) } // callback(unsigned long arg) without existing variable @change_callback_handle_cast_no_arg depends on change_timer_function_usage && !change_callback_handle_cast@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _origtype; identifier _origarg; type _handletype; @@ void _callback( -_origtype _origarg +struct timer_list *t ) { + _handletype *_origarg = from_timer(_origarg, t, _timer); + ... when != _origarg - (_handletype *)_origarg + _origarg ... when != _origarg } // Avoid already converted callbacks. @match_callback_converted depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier t; @@ void _callback(struct timer_list *t) { ... } // callback(struct something *handle) @change_callback_handle_arg depends on change_timer_function_usage && !match_callback_converted && !change_callback_handle_cast && !change_callback_handle_cast_no_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; @@ void _callback( -_handletype *_handle +struct timer_list *t ) { + _handletype *_handle = from_timer(_handle, t, _timer); ... } // If change_callback_handle_arg ran on an empty function, remove // the added handler. @unchange_callback_handle_arg depends on change_timer_function_usage && change_callback_handle_arg@ identifier change_timer_function_usage._callback; identifier change_timer_function_usage._timer; type _handletype; identifier _handle; identifier t; @@ void _callback(struct timer_list *t) { - _handletype *_handle = from_timer(_handle, t, _timer); } // We only want to refactor the setup_timer() data argument if we've found // the matching callback. This undoes changes in change_timer_function_usage. @unchange_timer_function_usage depends on change_timer_function_usage && !change_callback_handle_cast && !change_callback_handle_cast_no_arg && !change_callback_handle_arg@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type change_timer_function_usage._cast_data; @@ ( -timer_setup(&_E->_timer, _callback, 0); +setup_timer(&_E->_timer, _callback, (_cast_data)_E); | -timer_setup(&_E._timer, _callback, 0); +setup_timer(&_E._timer, _callback, (_cast_data)&_E); ) // If we fixed a callback from a .function assignment, fix the // assignment cast now. @change_timer_function_assignment depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression change_timer_function_usage._E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_func; typedef TIMER_FUNC_TYPE; @@ ( _E->_timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -&_callback +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)_callback; +(TIMER_FUNC_TYPE)_callback ; | _E->_timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -&_callback; +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)_callback +(TIMER_FUNC_TYPE)_callback ; | _E._timer.function = -(_cast_func)&_callback +(TIMER_FUNC_TYPE)_callback ; ) // Sometimes timer functions are called directly. Replace matched args. @change_timer_function_calls depends on change_timer_function_usage && (change_callback_handle_cast || change_callback_handle_cast_no_arg || change_callback_handle_arg)@ expression _E; identifier change_timer_function_usage._timer; identifier change_timer_function_usage._callback; type _cast_data; @@ _callback( ( -(_cast_data)_E +&_E->_timer | -(_cast_data)&_E +&_E._timer | -_E +&_E->_timer ) ) // If a timer has been configured without a data argument, it can be // converted without regard to the callback argument, since it is unused. @match_timer_function_unused_data@ expression _E; identifier _timer; identifier _callback; @@ ( -setup_timer(&_E->_timer, _callback, 0); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0L); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E->_timer, _callback, 0UL); +timer_setup(&_E->_timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0L); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_E._timer, _callback, 0UL); +timer_setup(&_E._timer, _callback, 0); | -setup_timer(&_timer, _callback, 0); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0L); +timer_setup(&_timer, _callback, 0); | -setup_timer(&_timer, _callback, 0UL); +timer_setup(&_timer, _callback, 0); | -setup_timer(_timer, _callback, 0); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0L); +timer_setup(_timer, _callback, 0); | -setup_timer(_timer, _callback, 0UL); +timer_setup(_timer, _callback, 0); ) @change_callback_unused_data depends on match_timer_function_unused_data@ identifier match_timer_function_unused_data._callback; type _origtype; identifier _origarg; @@ void _callback( -_origtype _origarg +struct timer_list *unused ) { ... when != _origarg } Signed-off-by: Kees Cook <keescook@chromium.org>
2017-10-17 05:43:17 +08:00
struct fcoe_ctlr *fip = from_timer(fip, t, timer);
schedule_work(&fip->timer_work);
}
/**
* fcoe_ctlr_timer_work() - Worker thread function for timer work
* @work: Handle to a FCoE controller
*
* Ages FCFs. Triggers FCF selection if possible.
* Sends keep-alives and resets.
*/
static void fcoe_ctlr_timer_work(struct work_struct *work)
{
struct fcoe_ctlr *fip;
struct fc_lport *vport;
u8 *mac;
u8 reset = 0;
u8 send_ctlr_ka = 0;
u8 send_port_ka = 0;
struct fcoe_fcf *sel;
struct fcoe_fcf *fcf;
unsigned long next_timer;
fip = container_of(work, struct fcoe_ctlr, timer_work);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (fip->mode == FIP_MODE_VN2VN)
return fcoe_ctlr_vn_timeout(fip);
mutex_lock(&fip->ctlr_mutex);
if (fip->state == FIP_ST_DISABLED) {
mutex_unlock(&fip->ctlr_mutex);
return;
}
fcf = fip->sel_fcf;
next_timer = fcoe_ctlr_age_fcfs(fip);
sel = fip->sel_fcf;
if (!sel && fip->sel_time) {
if (time_after_eq(jiffies, fip->sel_time)) {
sel = fcoe_ctlr_select(fip);
fip->sel_time = 0;
} else if (time_after(next_timer, fip->sel_time))
next_timer = fip->sel_time;
}
if (sel && fip->flogi_req_send)
fcoe_ctlr_flogi_send(fip);
else if (!sel && fcf)
reset = 1;
if (sel && !sel->fd_flags) {
if (time_after_eq(jiffies, fip->ctlr_ka_time)) {
fip->ctlr_ka_time = jiffies + sel->fka_period;
send_ctlr_ka = 1;
}
if (time_after(next_timer, fip->ctlr_ka_time))
next_timer = fip->ctlr_ka_time;
if (time_after_eq(jiffies, fip->port_ka_time)) {
fip->port_ka_time = jiffies +
msecs_to_jiffies(FIP_VN_KA_PERIOD);
send_port_ka = 1;
}
if (time_after(next_timer, fip->port_ka_time))
next_timer = fip->port_ka_time;
}
if (!list_empty(&fip->fcfs))
mod_timer(&fip->timer, next_timer);
mutex_unlock(&fip->ctlr_mutex);
if (reset) {
fc_lport_reset(fip->lp);
/* restart things with a solicitation */
fcoe_ctlr_solicit(fip, NULL);
}
if (send_ctlr_ka)
fcoe_ctlr_send_keep_alive(fip, NULL, 0, fip->ctl_src_addr);
if (send_port_ka) {
mutex_lock(&fip->lp->lp_mutex);
mac = fip->get_src_addr(fip->lp);
fcoe_ctlr_send_keep_alive(fip, fip->lp, 1, mac);
list_for_each_entry(vport, &fip->lp->vports, list) {
mac = fip->get_src_addr(vport);
fcoe_ctlr_send_keep_alive(fip, vport, 1, mac);
}
mutex_unlock(&fip->lp->lp_mutex);
}
}
/**
* fcoe_ctlr_recv_work() - Worker thread function for receiving FIP frames
* @recv_work: Handle to a FCoE controller
*/
static void fcoe_ctlr_recv_work(struct work_struct *recv_work)
{
struct fcoe_ctlr *fip;
struct sk_buff *skb;
fip = container_of(recv_work, struct fcoe_ctlr, recv_work);
while ((skb = skb_dequeue(&fip->fip_recv_list)))
fcoe_ctlr_recv_handler(fip, skb);
}
/**
* fcoe_ctlr_recv_flogi() - Snoop pre-FIP receipt of FLOGI response
* @fip: The FCoE controller
* @fp: The FC frame to snoop
*
* Snoop potential response to FLOGI or even incoming FLOGI.
*
* The caller has checked that we are waiting for login as indicated
* by fip->flogi_oxid != FC_XID_UNKNOWN.
*
* The caller is responsible for freeing the frame.
* Fill in the granted_mac address.
*
* Return non-zero if the frame should not be delivered to libfc.
*/
int fcoe_ctlr_recv_flogi(struct fcoe_ctlr *fip, struct fc_lport *lport,
struct fc_frame *fp)
{
struct fc_frame_header *fh;
u8 op;
u8 *sa;
sa = eth_hdr(&fp->skb)->h_source;
fh = fc_frame_header_get(fp);
if (fh->fh_type != FC_TYPE_ELS)
return 0;
op = fc_frame_payload_op(fp);
if (op == ELS_LS_ACC && fh->fh_r_ctl == FC_RCTL_ELS_REP &&
fip->flogi_oxid == ntohs(fh->fh_ox_id)) {
mutex_lock(&fip->ctlr_mutex);
if (fip->state != FIP_ST_AUTO && fip->state != FIP_ST_NON_FIP) {
mutex_unlock(&fip->ctlr_mutex);
return -EINVAL;
}
fcoe_ctlr_set_state(fip, FIP_ST_NON_FIP);
LIBFCOE_FIP_DBG(fip,
"received FLOGI LS_ACC using non-FIP mode\n");
/*
* FLOGI accepted.
* If the src mac addr is FC_OUI-based, then we mark the
* address_mode flag to use FC_OUI-based Ethernet DA.
* Otherwise we use the FCoE gateway addr
*/
if (ether_addr_equal(sa, (u8[6])FC_FCOE_FLOGI_MAC)) {
fcoe_ctlr_map_dest(fip);
} else {
memcpy(fip->dest_addr, sa, ETH_ALEN);
fip->map_dest = 0;
}
fip->flogi_oxid = FC_XID_UNKNOWN;
mutex_unlock(&fip->ctlr_mutex);
fc_fcoe_set_mac(fr_cb(fp)->granted_mac, fh->fh_d_id);
} else if (op == ELS_FLOGI && fh->fh_r_ctl == FC_RCTL_ELS_REQ && sa) {
/*
* Save source MAC for point-to-point responses.
*/
mutex_lock(&fip->ctlr_mutex);
if (fip->state == FIP_ST_AUTO || fip->state == FIP_ST_NON_FIP) {
memcpy(fip->dest_addr, sa, ETH_ALEN);
fip->map_dest = 0;
if (fip->state == FIP_ST_AUTO)
LIBFCOE_FIP_DBG(fip, "received non-FIP FLOGI. "
"Setting non-FIP mode\n");
fcoe_ctlr_set_state(fip, FIP_ST_NON_FIP);
}
mutex_unlock(&fip->ctlr_mutex);
}
return 0;
}
EXPORT_SYMBOL(fcoe_ctlr_recv_flogi);
/**
* fcoe_wwn_from_mac() - Converts a 48-bit IEEE MAC address to a 64-bit FC WWN
* @mac: The MAC address to convert
* @scheme: The scheme to use when converting
* @port: The port indicator for converting
*
* Returns: u64 fc world wide name
*/
u64 fcoe_wwn_from_mac(unsigned char mac[MAX_ADDR_LEN],
unsigned int scheme, unsigned int port)
{
u64 wwn;
u64 host_mac;
/* The MAC is in NO, so flip only the low 48 bits */
host_mac = ((u64) mac[0] << 40) |
((u64) mac[1] << 32) |
((u64) mac[2] << 24) |
((u64) mac[3] << 16) |
((u64) mac[4] << 8) |
(u64) mac[5];
WARN_ON(host_mac >= (1ULL << 48));
wwn = host_mac | ((u64) scheme << 60);
switch (scheme) {
case 1:
WARN_ON(port != 0);
break;
case 2:
WARN_ON(port >= 0xfff);
wwn |= (u64) port << 48;
break;
default:
WARN_ON(1);
break;
}
return wwn;
}
EXPORT_SYMBOL_GPL(fcoe_wwn_from_mac);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
/**
* fcoe_ctlr_rport() - return the fcoe_rport for a given fc_rport_priv
* @rdata: libfc remote port
*/
static inline struct fcoe_rport *fcoe_ctlr_rport(struct fc_rport_priv *rdata)
{
return (struct fcoe_rport *)(rdata + 1);
}
/**
* fcoe_ctlr_vn_send() - Send a FIP VN2VN Probe Request or Reply.
* @fip: The FCoE controller
* @sub: sub-opcode for probe request, reply, or advertisement.
* @dest: The destination Ethernet MAC address
* @min_len: minimum size of the Ethernet payload to be sent
*/
static void fcoe_ctlr_vn_send(struct fcoe_ctlr *fip,
enum fip_vn2vn_subcode sub,
const u8 *dest, size_t min_len)
{
struct sk_buff *skb;
struct fip_vn2vn_probe_frame {
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
struct ethhdr eth;
struct fip_header fip;
struct fip_mac_desc mac;
struct fip_wwn_desc wwnn;
struct fip_vn_desc vn;
} __packed * frame;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
struct fip_fc4_feat *ff;
struct fip_size_desc *size;
u32 fcp_feat;
size_t len;
size_t dlen;
len = sizeof(*frame);
dlen = 0;
if (sub == FIP_SC_VN_CLAIM_NOTIFY || sub == FIP_SC_VN_CLAIM_REP) {
dlen = sizeof(struct fip_fc4_feat) +
sizeof(struct fip_size_desc);
len += dlen;
}
dlen += sizeof(frame->mac) + sizeof(frame->wwnn) + sizeof(frame->vn);
len = max(len, min_len + sizeof(struct ethhdr));
skb = dev_alloc_skb(len);
if (!skb)
return;
frame = (struct fip_vn2vn_probe_frame *)skb->data;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
memset(frame, 0, len);
memcpy(frame->eth.h_dest, dest, ETH_ALEN);
if (sub == FIP_SC_VN_BEACON) {
hton24(frame->eth.h_source, FIP_VN_FC_MAP);
hton24(frame->eth.h_source + 3, fip->port_id);
} else {
memcpy(frame->eth.h_source, fip->ctl_src_addr, ETH_ALEN);
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
frame->eth.h_proto = htons(ETH_P_FIP);
frame->fip.fip_ver = FIP_VER_ENCAPS(FIP_VER);
frame->fip.fip_op = htons(FIP_OP_VN2VN);
frame->fip.fip_subcode = sub;
frame->fip.fip_dl_len = htons(dlen / FIP_BPW);
frame->mac.fd_desc.fip_dtype = FIP_DT_MAC;
frame->mac.fd_desc.fip_dlen = sizeof(frame->mac) / FIP_BPW;
memcpy(frame->mac.fd_mac, fip->ctl_src_addr, ETH_ALEN);
frame->wwnn.fd_desc.fip_dtype = FIP_DT_NAME;
frame->wwnn.fd_desc.fip_dlen = sizeof(frame->wwnn) / FIP_BPW;
put_unaligned_be64(fip->lp->wwnn, &frame->wwnn.fd_wwn);
frame->vn.fd_desc.fip_dtype = FIP_DT_VN_ID;
frame->vn.fd_desc.fip_dlen = sizeof(frame->vn) / FIP_BPW;
hton24(frame->vn.fd_mac, FIP_VN_FC_MAP);
hton24(frame->vn.fd_mac + 3, fip->port_id);
hton24(frame->vn.fd_fc_id, fip->port_id);
put_unaligned_be64(fip->lp->wwpn, &frame->vn.fd_wwpn);
/*
* For claims, add FC-4 features.
* TBD: Add interface to get fc-4 types and features from libfc.
*/
if (sub == FIP_SC_VN_CLAIM_NOTIFY || sub == FIP_SC_VN_CLAIM_REP) {
ff = (struct fip_fc4_feat *)(frame + 1);
ff->fd_desc.fip_dtype = FIP_DT_FC4F;
ff->fd_desc.fip_dlen = sizeof(*ff) / FIP_BPW;
ff->fd_fts = fip->lp->fcts;
fcp_feat = 0;
if (fip->lp->service_params & FCP_SPPF_INIT_FCN)
fcp_feat |= FCP_FEAT_INIT;
if (fip->lp->service_params & FCP_SPPF_TARG_FCN)
fcp_feat |= FCP_FEAT_TARG;
fcp_feat <<= (FC_TYPE_FCP * 4) % 32;
ff->fd_ff.fd_feat[FC_TYPE_FCP * 4 / 32] = htonl(fcp_feat);
size = (struct fip_size_desc *)(ff + 1);
size->fd_desc.fip_dtype = FIP_DT_FCOE_SIZE;
size->fd_desc.fip_dlen = sizeof(*size) / FIP_BPW;
size->fd_size = htons(fcoe_ctlr_fcoe_size(fip));
}
skb_put(skb, len);
skb->protocol = htons(ETH_P_FIP);
skb->priority = fip->priority;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
skb_reset_mac_header(skb);
skb_reset_network_header(skb);
fip->send(fip, skb);
}
/**
* fcoe_ctlr_vn_rport_callback - Event handler for rport events.
* @lport: The lport which is receiving the event
* @rdata: remote port private data
* @event: The event that occurred
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
*
* Locking Note: The rport lock must not be held when calling this function.
*/
static void fcoe_ctlr_vn_rport_callback(struct fc_lport *lport,
struct fc_rport_priv *rdata,
enum fc_rport_event event)
{
struct fcoe_ctlr *fip = lport->disc.priv;
struct fcoe_rport *frport = fcoe_ctlr_rport(rdata);
LIBFCOE_FIP_DBG(fip, "vn_rport_callback %x event %d\n",
rdata->ids.port_id, event);
mutex_lock(&fip->ctlr_mutex);
switch (event) {
case RPORT_EV_READY:
frport->login_count = 0;
break;
case RPORT_EV_LOGO:
case RPORT_EV_FAILED:
case RPORT_EV_STOP:
frport->login_count++;
if (frport->login_count > FCOE_CTLR_VN2VN_LOGIN_LIMIT) {
LIBFCOE_FIP_DBG(fip,
"rport FLOGI limited port_id %6.6x\n",
rdata->ids.port_id);
fc_rport_logoff(rdata);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
}
break;
default:
break;
}
mutex_unlock(&fip->ctlr_mutex);
}
static struct fc_rport_operations fcoe_ctlr_vn_rport_ops = {
.event_callback = fcoe_ctlr_vn_rport_callback,
};
/**
* fcoe_ctlr_disc_stop_locked() - stop discovery in VN2VN mode
* @fip: The FCoE controller
*
* Called with ctlr_mutex held.
*/
static void fcoe_ctlr_disc_stop_locked(struct fc_lport *lport)
{
struct fc_rport_priv *rdata;
rcu_read_lock();
list_for_each_entry_rcu(rdata, &lport->disc.rports, peers) {
if (kref_get_unless_zero(&rdata->kref)) {
fc_rport_logoff(rdata);
kref_put(&rdata->kref, fc_rport_destroy);
}
}
rcu_read_unlock();
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
mutex_lock(&lport->disc.disc_mutex);
lport->disc.disc_callback = NULL;
mutex_unlock(&lport->disc.disc_mutex);
}
/**
* fcoe_ctlr_disc_stop() - stop discovery in VN2VN mode
* @fip: The FCoE controller
*
* Called through the local port template for discovery.
* Called without the ctlr_mutex held.
*/
static void fcoe_ctlr_disc_stop(struct fc_lport *lport)
{
struct fcoe_ctlr *fip = lport->disc.priv;
mutex_lock(&fip->ctlr_mutex);
fcoe_ctlr_disc_stop_locked(lport);
mutex_unlock(&fip->ctlr_mutex);
}
/**
* fcoe_ctlr_disc_stop_final() - stop discovery for shutdown in VN2VN mode
* @fip: The FCoE controller
*
* Called through the local port template for discovery.
* Called without the ctlr_mutex held.
*/
static void fcoe_ctlr_disc_stop_final(struct fc_lport *lport)
{
fcoe_ctlr_disc_stop(lport);
fc_rport_flush_queue();
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
synchronize_rcu();
}
/**
* fcoe_ctlr_vn_restart() - VN2VN probe restart with new port_id
* @fip: The FCoE controller
*
* Called with fcoe_ctlr lock held.
*/
static void fcoe_ctlr_vn_restart(struct fcoe_ctlr *fip)
{
unsigned long wait;
u32 port_id;
fcoe_ctlr_disc_stop_locked(fip->lp);
/*
* Get proposed port ID.
* If this is the first try after link up, use any previous port_id.
* If there was none, use the low bits of the port_name.
* On subsequent tries, get the next random one.
* Don't use reserved IDs, use another non-zero value, just as random.
*/
port_id = fip->port_id;
if (fip->probe_tries)
random32: rename random32 to prandom This renames all random32 functions to have 'prandom_' prefix as follows: void prandom_seed(u32 seed); /* rename from srandom32() */ u32 prandom_u32(void); /* rename from random32() */ void prandom_seed_state(struct rnd_state *state, u64 seed); /* rename from prandom32_seed() */ u32 prandom_u32_state(struct rnd_state *state); /* rename from prandom32() */ The purpose of this renaming is to prevent some kernel developers from assuming that prandom32() and random32() might imply that only prandom32() was the one using a pseudo-random number generator by prandom32's "p", and the result may be a very embarassing security exposure. This concern was expressed by Theodore Ts'o. And furthermore, I'm going to introduce new functions for getting the requested number of pseudo-random bytes. If I continue to use both prandom32 and random32 prefixes for these functions, the confusion is getting worse. As a result of this renaming, "prandom_" is the common prefix for pseudo-random number library. Currently, srandom32() and random32() are preserved because it is difficult to rename too many users at once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Robert Love <robert.w.love@intel.com> Cc: Michel Lespinasse <walken@google.com> Cc: Valdis Kletnieks <valdis.kletnieks@vt.edu> Cc: David Laight <david.laight@aculab.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Artem Bityutskiy <dedekind1@gmail.com> Cc: David Woodhouse <dwmw2@infradead.org> Cc: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-12-18 08:04:23 +08:00
port_id = prandom_u32_state(&fip->rnd_state) & 0xffff;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
else if (!port_id)
port_id = fip->lp->wwpn & 0xffff;
if (!port_id || port_id == 0xffff)
port_id = 1;
fip->port_id = port_id;
if (fip->probe_tries < FIP_VN_RLIM_COUNT) {
fip->probe_tries++;
wait = prandom_u32() % FIP_VN_PROBE_WAIT;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
} else
wait = FIP_VN_RLIM_INT;
mod_timer(&fip->timer, jiffies + msecs_to_jiffies(wait));
fcoe_ctlr_set_state(fip, FIP_ST_VNMP_START);
}
/**
* fcoe_ctlr_vn_start() - Start in VN2VN mode
* @fip: The FCoE controller
*
* Called with fcoe_ctlr lock held.
*/
static void fcoe_ctlr_vn_start(struct fcoe_ctlr *fip)
{
fip->probe_tries = 0;
random32: rename random32 to prandom This renames all random32 functions to have 'prandom_' prefix as follows: void prandom_seed(u32 seed); /* rename from srandom32() */ u32 prandom_u32(void); /* rename from random32() */ void prandom_seed_state(struct rnd_state *state, u64 seed); /* rename from prandom32_seed() */ u32 prandom_u32_state(struct rnd_state *state); /* rename from prandom32() */ The purpose of this renaming is to prevent some kernel developers from assuming that prandom32() and random32() might imply that only prandom32() was the one using a pseudo-random number generator by prandom32's "p", and the result may be a very embarassing security exposure. This concern was expressed by Theodore Ts'o. And furthermore, I'm going to introduce new functions for getting the requested number of pseudo-random bytes. If I continue to use both prandom32 and random32 prefixes for these functions, the confusion is getting worse. As a result of this renaming, "prandom_" is the common prefix for pseudo-random number library. Currently, srandom32() and random32() are preserved because it is difficult to rename too many users at once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Robert Love <robert.w.love@intel.com> Cc: Michel Lespinasse <walken@google.com> Cc: Valdis Kletnieks <valdis.kletnieks@vt.edu> Cc: David Laight <david.laight@aculab.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Artem Bityutskiy <dedekind1@gmail.com> Cc: David Woodhouse <dwmw2@infradead.org> Cc: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-12-18 08:04:23 +08:00
prandom_seed_state(&fip->rnd_state, fip->lp->wwpn);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_restart(fip);
}
/**
* fcoe_ctlr_vn_parse - parse probe request or response
* @fip: The FCoE controller
* @skb: incoming packet
* @rdata: buffer for resulting parsed VN entry plus fcoe_rport
*
* Returns non-zero error number on error.
* Does not consume the packet.
*/
static int fcoe_ctlr_vn_parse(struct fcoe_ctlr *fip,
struct sk_buff *skb,
struct fc_rport_priv *rdata)
{
struct fip_header *fiph;
struct fip_desc *desc = NULL;
struct fip_mac_desc *macd = NULL;
struct fip_wwn_desc *wwn = NULL;
struct fip_vn_desc *vn = NULL;
struct fip_size_desc *size = NULL;
struct fcoe_rport *frport;
size_t rlen;
size_t dlen;
u32 desc_mask = 0;
u32 dtype;
u8 sub;
memset(rdata, 0, sizeof(*rdata) + sizeof(*frport));
frport = fcoe_ctlr_rport(rdata);
fiph = (struct fip_header *)skb->data;
frport->flags = ntohs(fiph->fip_flags);
sub = fiph->fip_subcode;
switch (sub) {
case FIP_SC_VN_PROBE_REQ:
case FIP_SC_VN_PROBE_REP:
case FIP_SC_VN_BEACON:
desc_mask = BIT(FIP_DT_MAC) | BIT(FIP_DT_NAME) |
BIT(FIP_DT_VN_ID);
break;
case FIP_SC_VN_CLAIM_NOTIFY:
case FIP_SC_VN_CLAIM_REP:
desc_mask = BIT(FIP_DT_MAC) | BIT(FIP_DT_NAME) |
BIT(FIP_DT_VN_ID) | BIT(FIP_DT_FC4F) |
BIT(FIP_DT_FCOE_SIZE);
break;
default:
LIBFCOE_FIP_DBG(fip, "vn_parse unknown subcode %u\n", sub);
return -EINVAL;
}
rlen = ntohs(fiph->fip_dl_len) * 4;
if (rlen + sizeof(*fiph) > skb->len)
return -EINVAL;
desc = (struct fip_desc *)(fiph + 1);
while (rlen > 0) {
dlen = desc->fip_dlen * FIP_BPW;
if (dlen < sizeof(*desc) || dlen > rlen)
return -EINVAL;
dtype = desc->fip_dtype;
if (dtype < 32) {
if (!(desc_mask & BIT(dtype))) {
LIBFCOE_FIP_DBG(fip,
"unexpected or duplicated desc "
"desc type %u in "
"FIP VN2VN subtype %u\n",
dtype, sub);
return -EINVAL;
}
desc_mask &= ~BIT(dtype);
}
switch (dtype) {
case FIP_DT_MAC:
if (dlen != sizeof(struct fip_mac_desc))
goto len_err;
macd = (struct fip_mac_desc *)desc;
if (!is_valid_ether_addr(macd->fd_mac)) {
LIBFCOE_FIP_DBG(fip,
"Invalid MAC addr %pM in FIP VN2VN\n",
macd->fd_mac);
return -EINVAL;
}
memcpy(frport->enode_mac, macd->fd_mac, ETH_ALEN);
break;
case FIP_DT_NAME:
if (dlen != sizeof(struct fip_wwn_desc))
goto len_err;
wwn = (struct fip_wwn_desc *)desc;
rdata->ids.node_name = get_unaligned_be64(&wwn->fd_wwn);
break;
case FIP_DT_VN_ID:
if (dlen != sizeof(struct fip_vn_desc))
goto len_err;
vn = (struct fip_vn_desc *)desc;
memcpy(frport->vn_mac, vn->fd_mac, ETH_ALEN);
rdata->ids.port_id = ntoh24(vn->fd_fc_id);
rdata->ids.port_name = get_unaligned_be64(&vn->fd_wwpn);
break;
case FIP_DT_FC4F:
if (dlen != sizeof(struct fip_fc4_feat))
goto len_err;
break;
case FIP_DT_FCOE_SIZE:
if (dlen != sizeof(struct fip_size_desc))
goto len_err;
size = (struct fip_size_desc *)desc;
frport->fcoe_len = ntohs(size->fd_size);
break;
default:
LIBFCOE_FIP_DBG(fip, "unexpected descriptor type %x "
"in FIP probe\n", dtype);
/* standard says ignore unknown descriptors >= 128 */
if (dtype < FIP_DT_NON_CRITICAL)
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
return -EINVAL;
break;
}
desc = (struct fip_desc *)((char *)desc + dlen);
rlen -= dlen;
}
return 0;
len_err:
LIBFCOE_FIP_DBG(fip, "FIP length error in descriptor type %x len %zu\n",
dtype, dlen);
return -EINVAL;
}
/**
* fcoe_ctlr_vn_send_claim() - send multicast FIP VN2VN Claim Notification.
* @fip: The FCoE controller
*
* Called with ctlr_mutex held.
*/
static void fcoe_ctlr_vn_send_claim(struct fcoe_ctlr *fip)
{
fcoe_ctlr_vn_send(fip, FIP_SC_VN_CLAIM_NOTIFY, fcoe_all_vn2vn, 0);
fip->sol_time = jiffies;
}
/**
* fcoe_ctlr_vn_probe_req() - handle incoming VN2VN probe request.
* @fip: The FCoE controller
* @rdata: parsed remote port with frport from the probe request
*
* Called with ctlr_mutex held.
*/
static void fcoe_ctlr_vn_probe_req(struct fcoe_ctlr *fip,
struct fc_rport_priv *rdata)
{
struct fcoe_rport *frport = fcoe_ctlr_rport(rdata);
if (rdata->ids.port_id != fip->port_id)
return;
switch (fip->state) {
case FIP_ST_VNMP_CLAIM:
case FIP_ST_VNMP_UP:
LIBFCOE_FIP_DBG(fip, "vn_probe_req: send reply, state %x\n",
fip->state);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send(fip, FIP_SC_VN_PROBE_REP,
frport->enode_mac, 0);
break;
case FIP_ST_VNMP_PROBE1:
case FIP_ST_VNMP_PROBE2:
/*
* Decide whether to reply to the Probe.
* Our selected address is never a "recorded" one, so
* only reply if our WWPN is greater and the
* Probe's REC bit is not set.
* If we don't reply, we will change our address.
*/
if (fip->lp->wwpn > rdata->ids.port_name &&
!(frport->flags & FIP_FL_REC_OR_P2P)) {
LIBFCOE_FIP_DBG(fip, "vn_probe_req: "
"port_id collision\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send(fip, FIP_SC_VN_PROBE_REP,
frport->enode_mac, 0);
break;
}
/* fall through */
case FIP_ST_VNMP_START:
LIBFCOE_FIP_DBG(fip, "vn_probe_req: "
"restart VN2VN negotiation\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_restart(fip);
break;
default:
LIBFCOE_FIP_DBG(fip, "vn_probe_req: ignore state %x\n",
fip->state);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
break;
}
}
/**
* fcoe_ctlr_vn_probe_reply() - handle incoming VN2VN probe reply.
* @fip: The FCoE controller
* @rdata: parsed remote port with frport from the probe request
*
* Called with ctlr_mutex held.
*/
static void fcoe_ctlr_vn_probe_reply(struct fcoe_ctlr *fip,
struct fc_rport_priv *rdata)
{
if (rdata->ids.port_id != fip->port_id)
return;
switch (fip->state) {
case FIP_ST_VNMP_START:
case FIP_ST_VNMP_PROBE1:
case FIP_ST_VNMP_PROBE2:
case FIP_ST_VNMP_CLAIM:
LIBFCOE_FIP_DBG(fip, "vn_probe_reply: restart state %x\n",
fip->state);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_restart(fip);
break;
case FIP_ST_VNMP_UP:
LIBFCOE_FIP_DBG(fip, "vn_probe_reply: send claim notify\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send_claim(fip);
break;
default:
break;
}
}
/**
* fcoe_ctlr_vn_add() - Add a VN2VN entry to the list, based on a claim reply.
* @fip: The FCoE controller
* @new: newly-parsed remote port with frport as a template for new rdata
*
* Called with ctlr_mutex held.
*/
static void fcoe_ctlr_vn_add(struct fcoe_ctlr *fip, struct fc_rport_priv *new)
{
struct fc_lport *lport = fip->lp;
struct fc_rport_priv *rdata;
struct fc_rport_identifiers *ids;
struct fcoe_rport *frport;
u32 port_id;
port_id = new->ids.port_id;
if (port_id == fip->port_id)
return;
mutex_lock(&lport->disc.disc_mutex);
rdata = fc_rport_create(lport, port_id);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (!rdata) {
mutex_unlock(&lport->disc.disc_mutex);
return;
}
mutex_lock(&rdata->rp_mutex);
mutex_unlock(&lport->disc.disc_mutex);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
rdata->ops = &fcoe_ctlr_vn_rport_ops;
rdata->disc_id = lport->disc.disc_id;
ids = &rdata->ids;
if ((ids->port_name != -1 && ids->port_name != new->ids.port_name) ||
(ids->node_name != -1 && ids->node_name != new->ids.node_name)) {
mutex_unlock(&rdata->rp_mutex);
LIBFCOE_FIP_DBG(fip, "vn_add rport logoff %6.6x\n", port_id);
fc_rport_logoff(rdata);
mutex_lock(&rdata->rp_mutex);
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
ids->port_name = new->ids.port_name;
ids->node_name = new->ids.node_name;
mutex_unlock(&rdata->rp_mutex);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
frport = fcoe_ctlr_rport(rdata);
LIBFCOE_FIP_DBG(fip, "vn_add rport %6.6x %s state %d\n",
port_id, frport->fcoe_len ? "old" : "new",
rdata->rp_state);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
*frport = *fcoe_ctlr_rport(new);
frport->time = 0;
}
/**
* fcoe_ctlr_vn_lookup() - Find VN remote port's MAC address
* @fip: The FCoE controller
* @port_id: The port_id of the remote VN_node
* @mac: buffer which will hold the VN_NODE destination MAC address, if found.
*
* Returns non-zero error if no remote port found.
*/
static int fcoe_ctlr_vn_lookup(struct fcoe_ctlr *fip, u32 port_id, u8 *mac)
{
struct fc_lport *lport = fip->lp;
struct fc_rport_priv *rdata;
struct fcoe_rport *frport;
int ret = -1;
rdata = fc_rport_lookup(lport, port_id);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (rdata) {
frport = fcoe_ctlr_rport(rdata);
memcpy(mac, frport->enode_mac, ETH_ALEN);
ret = 0;
kref_put(&rdata->kref, fc_rport_destroy);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
}
return ret;
}
/**
* fcoe_ctlr_vn_claim_notify() - handle received FIP VN2VN Claim Notification
* @fip: The FCoE controller
* @new: newly-parsed remote port with frport as a template for new rdata
*
* Called with ctlr_mutex held.
*/
static void fcoe_ctlr_vn_claim_notify(struct fcoe_ctlr *fip,
struct fc_rport_priv *new)
{
struct fcoe_rport *frport = fcoe_ctlr_rport(new);
if (frport->flags & FIP_FL_REC_OR_P2P) {
LIBFCOE_FIP_DBG(fip, "send probe req for P2P/REC\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send(fip, FIP_SC_VN_PROBE_REQ, fcoe_all_vn2vn, 0);
return;
}
switch (fip->state) {
case FIP_ST_VNMP_START:
case FIP_ST_VNMP_PROBE1:
case FIP_ST_VNMP_PROBE2:
if (new->ids.port_id == fip->port_id) {
LIBFCOE_FIP_DBG(fip, "vn_claim_notify: "
"restart, state %d\n",
fip->state);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_restart(fip);
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
break;
case FIP_ST_VNMP_CLAIM:
case FIP_ST_VNMP_UP:
if (new->ids.port_id == fip->port_id) {
if (new->ids.port_name > fip->lp->wwpn) {
LIBFCOE_FIP_DBG(fip, "vn_claim_notify: "
"restart, port_id collision\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_restart(fip);
break;
}
LIBFCOE_FIP_DBG(fip, "vn_claim_notify: "
"send claim notify\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send_claim(fip);
break;
}
LIBFCOE_FIP_DBG(fip, "vn_claim_notify: send reply to %x\n",
new->ids.port_id);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send(fip, FIP_SC_VN_CLAIM_REP, frport->enode_mac,
min((u32)frport->fcoe_len,
fcoe_ctlr_fcoe_size(fip)));
fcoe_ctlr_vn_add(fip, new);
break;
default:
LIBFCOE_FIP_DBG(fip, "vn_claim_notify: "
"ignoring claim from %x\n", new->ids.port_id);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
break;
}
}
/**
* fcoe_ctlr_vn_claim_resp() - handle received Claim Response
* @fip: The FCoE controller that received the frame
* @new: newly-parsed remote port with frport from the Claim Response
*
* Called with ctlr_mutex held.
*/
static void fcoe_ctlr_vn_claim_resp(struct fcoe_ctlr *fip,
struct fc_rport_priv *new)
{
LIBFCOE_FIP_DBG(fip, "claim resp from from rport %x - state %s\n",
new->ids.port_id, fcoe_ctlr_state(fip->state));
if (fip->state == FIP_ST_VNMP_UP || fip->state == FIP_ST_VNMP_CLAIM)
fcoe_ctlr_vn_add(fip, new);
}
/**
* fcoe_ctlr_vn_beacon() - handle received beacon.
* @fip: The FCoE controller that received the frame
* @new: newly-parsed remote port with frport from the Beacon
*
* Called with ctlr_mutex held.
*/
static void fcoe_ctlr_vn_beacon(struct fcoe_ctlr *fip,
struct fc_rport_priv *new)
{
struct fc_lport *lport = fip->lp;
struct fc_rport_priv *rdata;
struct fcoe_rport *frport;
frport = fcoe_ctlr_rport(new);
if (frport->flags & FIP_FL_REC_OR_P2P) {
LIBFCOE_FIP_DBG(fip, "p2p beacon while in vn2vn mode\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send(fip, FIP_SC_VN_PROBE_REQ, fcoe_all_vn2vn, 0);
return;
}
rdata = fc_rport_lookup(lport, new->ids.port_id);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (rdata) {
if (rdata->ids.node_name == new->ids.node_name &&
rdata->ids.port_name == new->ids.port_name) {
frport = fcoe_ctlr_rport(rdata);
LIBFCOE_FIP_DBG(fip, "beacon from rport %x\n",
rdata->ids.port_id);
if (!frport->time && fip->state == FIP_ST_VNMP_UP) {
LIBFCOE_FIP_DBG(fip, "beacon expired "
"for rport %x\n",
rdata->ids.port_id);
fc_rport_login(rdata);
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
frport->time = jiffies;
}
kref_put(&rdata->kref, fc_rport_destroy);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
return;
}
if (fip->state != FIP_ST_VNMP_UP)
return;
/*
* Beacon from a new neighbor.
* Send a claim notify if one hasn't been sent recently.
* Don't add the neighbor yet.
*/
LIBFCOE_FIP_DBG(fip, "beacon from new rport %x. sending claim notify\n",
new->ids.port_id);
if (time_after(jiffies,
fip->sol_time + msecs_to_jiffies(FIP_VN_ANN_WAIT)))
fcoe_ctlr_vn_send_claim(fip);
}
/**
* fcoe_ctlr_vn_age() - Check for VN_ports without recent beacons
* @fip: The FCoE controller
*
* Called with ctlr_mutex held.
* Called only in state FIP_ST_VNMP_UP.
* Returns the soonest time for next age-out or a time far in the future.
*/
static unsigned long fcoe_ctlr_vn_age(struct fcoe_ctlr *fip)
{
struct fc_lport *lport = fip->lp;
struct fc_rport_priv *rdata;
struct fcoe_rport *frport;
unsigned long next_time;
unsigned long deadline;
next_time = jiffies + msecs_to_jiffies(FIP_VN_BEACON_INT * 10);
rcu_read_lock();
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
list_for_each_entry_rcu(rdata, &lport->disc.rports, peers) {
if (!kref_get_unless_zero(&rdata->kref))
continue;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
frport = fcoe_ctlr_rport(rdata);
if (!frport->time) {
kref_put(&rdata->kref, fc_rport_destroy);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
continue;
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
deadline = frport->time +
msecs_to_jiffies(FIP_VN_BEACON_INT * 25 / 10);
if (time_after_eq(jiffies, deadline)) {
frport->time = 0;
LIBFCOE_FIP_DBG(fip,
"port %16.16llx fc_id %6.6x beacon expired\n",
rdata->ids.port_name, rdata->ids.port_id);
fc_rport_logoff(rdata);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
} else if (time_before(deadline, next_time))
next_time = deadline;
kref_put(&rdata->kref, fc_rport_destroy);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
}
rcu_read_unlock();
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
return next_time;
}
/**
* fcoe_ctlr_vn_recv() - Receive a FIP frame
* @fip: The FCoE controller that received the frame
* @skb: The received FIP frame
*
* Returns non-zero if the frame is dropped.
* Always consumes the frame.
*/
static int fcoe_ctlr_vn_recv(struct fcoe_ctlr *fip, struct sk_buff *skb)
{
struct fip_header *fiph;
enum fip_vn2vn_subcode sub;
struct {
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
struct fc_rport_priv rdata;
struct fcoe_rport frport;
} buf;
int rc, vlan_id = 0;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fiph = (struct fip_header *)skb->data;
sub = fiph->fip_subcode;
if (fip->lp->vlan)
vlan_id = skb_vlan_tag_get_id(skb);
if (vlan_id && vlan_id != fip->lp->vlan) {
LIBFCOE_FIP_DBG(fip, "vn_recv drop frame sub %x vlan %d\n",
sub, vlan_id);
rc = -EAGAIN;
goto drop;
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
rc = fcoe_ctlr_vn_parse(fip, skb, &buf.rdata);
if (rc) {
LIBFCOE_FIP_DBG(fip, "vn_recv vn_parse error %d\n", rc);
goto drop;
}
mutex_lock(&fip->ctlr_mutex);
switch (sub) {
case FIP_SC_VN_PROBE_REQ:
fcoe_ctlr_vn_probe_req(fip, &buf.rdata);
break;
case FIP_SC_VN_PROBE_REP:
fcoe_ctlr_vn_probe_reply(fip, &buf.rdata);
break;
case FIP_SC_VN_CLAIM_NOTIFY:
fcoe_ctlr_vn_claim_notify(fip, &buf.rdata);
break;
case FIP_SC_VN_CLAIM_REP:
fcoe_ctlr_vn_claim_resp(fip, &buf.rdata);
break;
case FIP_SC_VN_BEACON:
fcoe_ctlr_vn_beacon(fip, &buf.rdata);
break;
default:
LIBFCOE_FIP_DBG(fip, "vn_recv unknown subcode %d\n", sub);
rc = -1;
break;
}
mutex_unlock(&fip->ctlr_mutex);
drop:
kfree_skb(skb);
return rc;
}
/**
* fcoe_ctlr_vlan_parse - parse vlan discovery request or response
* @fip: The FCoE controller
* @skb: incoming packet
* @rdata: buffer for resulting parsed VLAN entry plus fcoe_rport
*
* Returns non-zero error number on error.
* Does not consume the packet.
*/
static int fcoe_ctlr_vlan_parse(struct fcoe_ctlr *fip,
struct sk_buff *skb,
struct fc_rport_priv *rdata)
{
struct fip_header *fiph;
struct fip_desc *desc = NULL;
struct fip_mac_desc *macd = NULL;
struct fip_wwn_desc *wwn = NULL;
struct fcoe_rport *frport;
size_t rlen;
size_t dlen;
u32 desc_mask = 0;
u32 dtype;
u8 sub;
memset(rdata, 0, sizeof(*rdata) + sizeof(*frport));
frport = fcoe_ctlr_rport(rdata);
fiph = (struct fip_header *)skb->data;
frport->flags = ntohs(fiph->fip_flags);
sub = fiph->fip_subcode;
switch (sub) {
case FIP_SC_VL_REQ:
desc_mask = BIT(FIP_DT_MAC) | BIT(FIP_DT_NAME);
break;
default:
LIBFCOE_FIP_DBG(fip, "vn_parse unknown subcode %u\n", sub);
return -EINVAL;
}
rlen = ntohs(fiph->fip_dl_len) * 4;
if (rlen + sizeof(*fiph) > skb->len)
return -EINVAL;
desc = (struct fip_desc *)(fiph + 1);
while (rlen > 0) {
dlen = desc->fip_dlen * FIP_BPW;
if (dlen < sizeof(*desc) || dlen > rlen)
return -EINVAL;
dtype = desc->fip_dtype;
if (dtype < 32) {
if (!(desc_mask & BIT(dtype))) {
LIBFCOE_FIP_DBG(fip,
"unexpected or duplicated desc "
"desc type %u in "
"FIP VN2VN subtype %u\n",
dtype, sub);
return -EINVAL;
}
desc_mask &= ~BIT(dtype);
}
switch (dtype) {
case FIP_DT_MAC:
if (dlen != sizeof(struct fip_mac_desc))
goto len_err;
macd = (struct fip_mac_desc *)desc;
if (!is_valid_ether_addr(macd->fd_mac)) {
LIBFCOE_FIP_DBG(fip,
"Invalid MAC addr %pM in FIP VN2VN\n",
macd->fd_mac);
return -EINVAL;
}
memcpy(frport->enode_mac, macd->fd_mac, ETH_ALEN);
break;
case FIP_DT_NAME:
if (dlen != sizeof(struct fip_wwn_desc))
goto len_err;
wwn = (struct fip_wwn_desc *)desc;
rdata->ids.node_name = get_unaligned_be64(&wwn->fd_wwn);
break;
default:
LIBFCOE_FIP_DBG(fip, "unexpected descriptor type %x "
"in FIP probe\n", dtype);
/* standard says ignore unknown descriptors >= 128 */
if (dtype < FIP_DT_NON_CRITICAL)
return -EINVAL;
break;
}
desc = (struct fip_desc *)((char *)desc + dlen);
rlen -= dlen;
}
return 0;
len_err:
LIBFCOE_FIP_DBG(fip, "FIP length error in descriptor type %x len %zu\n",
dtype, dlen);
return -EINVAL;
}
/**
* fcoe_ctlr_vlan_send() - Send a FIP VLAN Notification
* @fip: The FCoE controller
* @sub: sub-opcode for vlan notification or vn2vn vlan notification
* @dest: The destination Ethernet MAC address
* @min_len: minimum size of the Ethernet payload to be sent
*/
static void fcoe_ctlr_vlan_send(struct fcoe_ctlr *fip,
enum fip_vlan_subcode sub,
const u8 *dest)
{
struct sk_buff *skb;
struct fip_vlan_notify_frame {
struct ethhdr eth;
struct fip_header fip;
struct fip_mac_desc mac;
struct fip_vlan_desc vlan;
} __packed * frame;
size_t len;
size_t dlen;
len = sizeof(*frame);
dlen = sizeof(frame->mac) + sizeof(frame->vlan);
len = max(len, sizeof(struct ethhdr));
skb = dev_alloc_skb(len);
if (!skb)
return;
LIBFCOE_FIP_DBG(fip, "fip %s vlan notification, vlan %d\n",
fip->mode == FIP_MODE_VN2VN ? "vn2vn" : "fcf",
fip->lp->vlan);
frame = (struct fip_vlan_notify_frame *)skb->data;
memset(frame, 0, len);
memcpy(frame->eth.h_dest, dest, ETH_ALEN);
memcpy(frame->eth.h_source, fip->ctl_src_addr, ETH_ALEN);
frame->eth.h_proto = htons(ETH_P_FIP);
frame->fip.fip_ver = FIP_VER_ENCAPS(FIP_VER);
frame->fip.fip_op = htons(FIP_OP_VLAN);
frame->fip.fip_subcode = sub;
frame->fip.fip_dl_len = htons(dlen / FIP_BPW);
frame->mac.fd_desc.fip_dtype = FIP_DT_MAC;
frame->mac.fd_desc.fip_dlen = sizeof(frame->mac) / FIP_BPW;
memcpy(frame->mac.fd_mac, fip->ctl_src_addr, ETH_ALEN);
frame->vlan.fd_desc.fip_dtype = FIP_DT_VLAN;
frame->vlan.fd_desc.fip_dlen = sizeof(frame->vlan) / FIP_BPW;
put_unaligned_be16(fip->lp->vlan, &frame->vlan.fd_vlan);
skb_put(skb, len);
skb->protocol = htons(ETH_P_FIP);
skb->priority = fip->priority;
skb_reset_mac_header(skb);
skb_reset_network_header(skb);
fip->send(fip, skb);
}
/**
* fcoe_ctlr_vlan_disk_reply() - send FIP VLAN Discovery Notification.
* @fip: The FCoE controller
*
* Called with ctlr_mutex held.
*/
static void fcoe_ctlr_vlan_disc_reply(struct fcoe_ctlr *fip,
struct fc_rport_priv *rdata)
{
struct fcoe_rport *frport = fcoe_ctlr_rport(rdata);
enum fip_vlan_subcode sub = FIP_SC_VL_NOTE;
if (fip->mode == FIP_MODE_VN2VN)
sub = FIP_SC_VL_VN2VN_NOTE;
fcoe_ctlr_vlan_send(fip, sub, frport->enode_mac);
}
/**
* fcoe_ctlr_vlan_recv - vlan request receive handler for VN2VN mode.
* @lport: The local port
* @fp: The received frame
*
*/
static int fcoe_ctlr_vlan_recv(struct fcoe_ctlr *fip, struct sk_buff *skb)
{
struct fip_header *fiph;
enum fip_vlan_subcode sub;
struct {
struct fc_rport_priv rdata;
struct fcoe_rport frport;
} buf;
int rc;
fiph = (struct fip_header *)skb->data;
sub = fiph->fip_subcode;
rc = fcoe_ctlr_vlan_parse(fip, skb, &buf.rdata);
if (rc) {
LIBFCOE_FIP_DBG(fip, "vlan_recv vlan_parse error %d\n", rc);
goto drop;
}
mutex_lock(&fip->ctlr_mutex);
if (sub == FIP_SC_VL_REQ)
fcoe_ctlr_vlan_disc_reply(fip, &buf.rdata);
mutex_unlock(&fip->ctlr_mutex);
drop:
kfree_skb(skb);
return rc;
}
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
/**
* fcoe_ctlr_disc_recv - discovery receive handler for VN2VN mode.
* @lport: The local port
* @fp: The received frame
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
*
* This should never be called since we don't see RSCNs or other
* fabric-generated ELSes.
*/
static void fcoe_ctlr_disc_recv(struct fc_lport *lport, struct fc_frame *fp)
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
{
struct fc_seq_els_data rjt_data;
rjt_data.reason = ELS_RJT_UNSUP;
rjt_data.explan = ELS_EXPL_NONE;
fc_seq_els_rsp_send(fp, ELS_LS_RJT, &rjt_data);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fc_frame_free(fp);
}
/**
* fcoe_ctlr_disc_recv - start discovery for VN2VN mode.
* @fip: The FCoE controller
*
* This sets a flag indicating that remote ports should be created
* and started for the peers we discover. We use the disc_callback
* pointer as that flag. Peers already discovered are created here.
*
* The lport lock is held during this call. The callback must be done
* later, without holding either the lport or discovery locks.
* The fcoe_ctlr lock may also be held during this call.
*/
static void fcoe_ctlr_disc_start(void (*callback)(struct fc_lport *,
enum fc_disc_event),
struct fc_lport *lport)
{
struct fc_disc *disc = &lport->disc;
struct fcoe_ctlr *fip = disc->priv;
mutex_lock(&disc->disc_mutex);
disc->disc_callback = callback;
disc->disc_id = (disc->disc_id + 2) | 1;
disc->pending = 1;
schedule_work(&fip->timer_work);
mutex_unlock(&disc->disc_mutex);
}
/**
* fcoe_ctlr_vn_disc() - report FIP VN_port discovery results after claim state.
* @fip: The FCoE controller
*
* Starts the FLOGI and PLOGI login process to each discovered rport for which
* we've received at least one beacon.
* Performs the discovery complete callback.
*/
static void fcoe_ctlr_vn_disc(struct fcoe_ctlr *fip)
{
struct fc_lport *lport = fip->lp;
struct fc_disc *disc = &lport->disc;
struct fc_rport_priv *rdata;
struct fcoe_rport *frport;
void (*callback)(struct fc_lport *, enum fc_disc_event);
mutex_lock(&disc->disc_mutex);
callback = disc->pending ? disc->disc_callback : NULL;
disc->pending = 0;
mutex_unlock(&disc->disc_mutex);
rcu_read_lock();
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
list_for_each_entry_rcu(rdata, &disc->rports, peers) {
if (!kref_get_unless_zero(&rdata->kref))
continue;
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
frport = fcoe_ctlr_rport(rdata);
if (frport->time)
fc_rport_login(rdata);
kref_put(&rdata->kref, fc_rport_destroy);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
}
rcu_read_unlock();
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (callback)
callback(lport, DISC_EV_SUCCESS);
}
/**
* fcoe_ctlr_vn_timeout - timer work function for VN2VN mode.
* @fip: The FCoE controller
*/
static void fcoe_ctlr_vn_timeout(struct fcoe_ctlr *fip)
{
unsigned long next_time;
u8 mac[ETH_ALEN];
u32 new_port_id = 0;
mutex_lock(&fip->ctlr_mutex);
switch (fip->state) {
case FIP_ST_VNMP_START:
fcoe_ctlr_set_state(fip, FIP_ST_VNMP_PROBE1);
LIBFCOE_FIP_DBG(fip, "vn_timeout: send 1st probe request\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send(fip, FIP_SC_VN_PROBE_REQ, fcoe_all_vn2vn, 0);
next_time = jiffies + msecs_to_jiffies(FIP_VN_PROBE_WAIT);
break;
case FIP_ST_VNMP_PROBE1:
fcoe_ctlr_set_state(fip, FIP_ST_VNMP_PROBE2);
LIBFCOE_FIP_DBG(fip, "vn_timeout: send 2nd probe request\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send(fip, FIP_SC_VN_PROBE_REQ, fcoe_all_vn2vn, 0);
next_time = jiffies + msecs_to_jiffies(FIP_VN_ANN_WAIT);
break;
case FIP_ST_VNMP_PROBE2:
fcoe_ctlr_set_state(fip, FIP_ST_VNMP_CLAIM);
new_port_id = fip->port_id;
hton24(mac, FIP_VN_FC_MAP);
hton24(mac + 3, new_port_id);
fcoe_ctlr_map_dest(fip);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fip->update_mac(fip->lp, mac);
LIBFCOE_FIP_DBG(fip, "vn_timeout: send claim notify\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send_claim(fip);
next_time = jiffies + msecs_to_jiffies(FIP_VN_ANN_WAIT);
break;
case FIP_ST_VNMP_CLAIM:
/*
* This may be invoked either by starting discovery so don't
* go to the next state unless it's been long enough.
*/
next_time = fip->sol_time + msecs_to_jiffies(FIP_VN_ANN_WAIT);
if (time_after_eq(jiffies, next_time)) {
fcoe_ctlr_set_state(fip, FIP_ST_VNMP_UP);
LIBFCOE_FIP_DBG(fip, "vn_timeout: send vn2vn beacon\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send(fip, FIP_SC_VN_BEACON,
fcoe_all_vn2vn, 0);
next_time = jiffies + msecs_to_jiffies(FIP_VN_ANN_WAIT);
fip->port_ka_time = next_time;
}
fcoe_ctlr_vn_disc(fip);
break;
case FIP_ST_VNMP_UP:
next_time = fcoe_ctlr_vn_age(fip);
if (time_after_eq(jiffies, fip->port_ka_time)) {
LIBFCOE_FIP_DBG(fip, "vn_timeout: send vn2vn beacon\n");
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
fcoe_ctlr_vn_send(fip, FIP_SC_VN_BEACON,
fcoe_all_vn2vn, 0);
fip->port_ka_time = jiffies +
msecs_to_jiffies(FIP_VN_BEACON_INT +
(prandom_u32() % FIP_VN_BEACON_FUZZ));
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
}
if (time_before(fip->port_ka_time, next_time))
next_time = fip->port_ka_time;
break;
case FIP_ST_LINK_WAIT:
goto unlock;
default:
WARN(1, "unexpected state %d\n", fip->state);
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
goto unlock;
}
mod_timer(&fip->timer, next_time);
unlock:
mutex_unlock(&fip->ctlr_mutex);
/* If port ID is new, notify local port after dropping ctlr_mutex */
if (new_port_id)
fc_lport_set_local_id(fip->lp, new_port_id);
}
/**
* fcoe_ctlr_mode_set() - Set or reset the ctlr's mode
* @lport: The local port to be (re)configured
* @fip: The FCoE controller whose mode is changing
* @fip_mode: The new fip mode
*
* Note that the we shouldn't be changing the libfc discovery settings
* (fc_disc_config) while an lport is going through the libfc state
* machine. The mode can only be changed when a fcoe_ctlr device is
* disabled, so that should ensure that this routine is only called
* when nothing is happening.
*/
static void fcoe_ctlr_mode_set(struct fc_lport *lport, struct fcoe_ctlr *fip,
enum fip_mode fip_mode)
{
void *priv;
WARN_ON(lport->state != LPORT_ST_RESET &&
lport->state != LPORT_ST_DISABLED);
if (fip_mode == FIP_MODE_VN2VN) {
lport->rport_priv_size = sizeof(struct fcoe_rport);
lport->point_to_multipoint = 1;
lport->tt.disc_recv_req = fcoe_ctlr_disc_recv;
lport->tt.disc_start = fcoe_ctlr_disc_start;
lport->tt.disc_stop = fcoe_ctlr_disc_stop;
lport->tt.disc_stop_final = fcoe_ctlr_disc_stop_final;
priv = fip;
} else {
lport->rport_priv_size = 0;
lport->point_to_multipoint = 0;
lport->tt.disc_recv_req = NULL;
lport->tt.disc_start = NULL;
lport->tt.disc_stop = NULL;
lport->tt.disc_stop_final = NULL;
priv = lport;
}
fc_disc_config(lport, priv);
}
/**
* fcoe_libfc_config() - Sets up libfc related properties for local port
* @lport: The local port to configure libfc for
* @fip: The FCoE controller in use by the local port
* @tt: The libfc function template
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
* @init_fcp: If non-zero, the FCP portion of libfc should be initialized
*
* Returns : 0 for success
*/
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
int fcoe_libfc_config(struct fc_lport *lport, struct fcoe_ctlr *fip,
const struct libfc_function_template *tt, int init_fcp)
{
/* Set the function pointers set by the LLDD */
memcpy(&lport->tt, tt, sizeof(*tt));
[SCSI] libfcoe: fcoe: fnic: add FIP VN2VN point-to-multipoint support The FC-BB-6 committee is proposing a new FIP usage model called VN_port to VN_port mode. It allows VN_ports to discover each other over a loss-free L2 Ethernet without any FCF or Fibre-channel fabric services. This is point-to-multipoint. There is also a variant of this called point-to-point which provides for making sure there is just one pair of ports operating over the Ethernet fabric. We add these new states: VNMP_START, _PROBE1, _PROBE2, _CLAIM, and _UP. These usually go quickly in that sequence. After waiting a random amount of time up to 100 ms in START, we select a pseudo-random proposed locally-unique port ID and send out probes in states PROBE1 and PROBE2, 100 ms apart. If no probe responses are heard, we proceed to CLAIM state 400 ms later and send a claim notification. We wait another 400 ms to receive claim responses, which give us a list of the other nodes on the network, including their FC-4 capabilities. After another 400 ms we go to VNMP_UP state and should start interoperating with any of the nodes for whic we receivec claim responses. More details are in the spec.j Add the new mode as FIP_MODE_VN2VN. The driver must specify explicitly that it wants to operate in this mode. There is no automatic detection between point-to-multipoint and fabric mode, and the local port initialization is affected, so it isn't anticipated that there will ever be any such automatic switchover. It may eventually be possible to have both fabric and VN2VN modes on the same L2 network, which may be done by two separate local VN_ports (lports). When in VN2VN mode, FIP replaces libfc's fabric-oriented discovery module with its own simple code that adds remote ports as they are discovered from incoming claim notifications and responses. These hooks are placed by fcoe_disc_init(). A linear list of discovered vn_ports is maintained under the fcoe_ctlr struct. It is expected to be short for now, and accessed infrequently. It is kept under RCU for lock-ordering reasons. The lport and/or rport mutexes may be held when we need to lookup a fcoe_vnport during an ELS send. Change fcoe_ctlr_encaps() to lookup the destination vn_port in the list of peers for the destination MAC address of the FIP-encapsulated frame. Add a new function fcoe_disc_init() to initialize just the discovery portion of libfcoe for VN2VN mode. Signed-off-by: Joe Eykholt <jeykholt@cisco.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2010-07-21 06:20:30 +08:00
if (init_fcp && fc_fcp_init(lport))
return -ENOMEM;
fc_exch_init(lport);
fc_elsct_init(lport);
fc_lport_init(lport);
fc_disc_init(lport);
fcoe_ctlr_mode_set(lport, fip, fip->mode);
return 0;
}
EXPORT_SYMBOL_GPL(fcoe_libfc_config);
void fcoe_fcf_get_selected(struct fcoe_fcf_device *fcf_dev)
{
struct fcoe_ctlr_device *ctlr_dev = fcoe_fcf_dev_to_ctlr_dev(fcf_dev);
struct fcoe_ctlr *fip = fcoe_ctlr_device_priv(ctlr_dev);
struct fcoe_fcf *fcf;
mutex_lock(&fip->ctlr_mutex);
mutex_lock(&ctlr_dev->lock);
fcf = fcoe_fcf_device_priv(fcf_dev);
if (fcf)
fcf_dev->selected = (fcf == fip->sel_fcf) ? 1 : 0;
else
fcf_dev->selected = 0;
mutex_unlock(&ctlr_dev->lock);
mutex_unlock(&fip->ctlr_mutex);
}
EXPORT_SYMBOL(fcoe_fcf_get_selected);
void fcoe_ctlr_set_fip_mode(struct fcoe_ctlr_device *ctlr_dev)
{
struct fcoe_ctlr *ctlr = fcoe_ctlr_device_priv(ctlr_dev);
struct fc_lport *lport = ctlr->lp;
mutex_lock(&ctlr->ctlr_mutex);
switch (ctlr_dev->mode) {
case FIP_CONN_TYPE_VN2VN:
ctlr->mode = FIP_MODE_VN2VN;
break;
case FIP_CONN_TYPE_FABRIC:
default:
ctlr->mode = FIP_MODE_FABRIC;
break;
}
mutex_unlock(&ctlr->ctlr_mutex);
fcoe_ctlr_mode_set(lport, ctlr, ctlr->mode);
}
EXPORT_SYMBOL(fcoe_ctlr_set_fip_mode);