When the SCSI midlayer is running error recovery, the low-level error
recovery in zfcp could be running and preventing the SCSI midlayer to
issue error recovery requests. To avoid unnecessary error recovery
escalation, wait for the zfcp erp to finish and retry if necessary.
While reworking the SCSI eh handlers, alsa cleanup the code and
simplify the interface from zfcp_scsi to the fsf layer.
Acked-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
For calls from zfcp erp, scsi_eh and sysfs switch the calls issuing
FSF requests to zfcp_fsf_req_sbal_get to wait for free SBALs.
Acked-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Only increment the req_id for successfully issued requests. This
avoids some confusion when debugging issued fsf requests.
Acked-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
The lock only needs to protect the softirq context called from qdio
against the userspace context called from sysfs. spin_lock and
spin_lock_bh is enough.
Acked-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Provide measurement data for the utilisation of the QDIO outbound queue.
The additional value allows to calculate an average queue utilisation
by looking at the deltas per time unit. Needed for capacity planning.
It is up to user space to handle wrap-arounds of the 64 bit value.
The new counter neatly complements the existing counter for queue full
conditions. That is why, both statistics counter have been integrated.
Signed-off-by: Martin Peschke <mpeschke@linux.vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Use the device pointer in zfcp_unit for tracking if we have a
registered SCSI device. With this approach, the flag
ZFCP_STATUS_UNIT_REGISTERED is only redundant and can be removed.
Acked-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
PORT_PHYS_CLOSING is only set and cleared, but not actually used
for status checking.
PORT_INVALID_WWPN is set when the GID_PN request does not return
a d_id for a remote port, e.g. when a remote port has been
unplugged. For this case, the d_id is zero. In the erp we can
check the d_id and use the normal escalation procedure that gives
up after three retries and remove the special case.
PORT_NO_WWPN is unused: Each port in the remote port list has a
valid wwpn. The WKA ports are now tracked outside the port
list. Remove the PORT_NO_WWPN flag, since this is no longer set
for any port.
Acked-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
The SUGGEST_* flags in the SCSI command result have been out of fashion
for a while and we don't actually use them in the error handling.
Remove the remaining occurrences.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Remove a message that was emitted for a port that could not initially
be opened. This is a rare case when the port discovery hits an
initiator port and only confuses the user with an initator port logged
in the message. Remove the whole special case: The failed "open port"
request triggers required follow-up actions anyway.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: Felix Beck <felix@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Add the support to send CT and ELS requests as unchained FSF requests. This is
required for older hardware and was somehow omitted during the cleanup of the
FSF layer. The req_count and resp_count attributes are unused, so remove them
instead of adding a special case for setting them. Also add debug data and a
warning, when the ct request hits a limit.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: Martin Petermann <martin@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
With the change to the dev_ message macros, the macro to get the busid
is only used in a few places. Remove it and directly get the dev_name
from the device.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
The port flag DID_DID indicates whether we know the current id of the
port. This is always set in parallel. Since the id 0 is invalid
(because the port id 0 is invalid) we can remove the DID_DID flag:
d_id of 0 indicates an invalid d_id != 0 is a valid one.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: Felix Beck <felix@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Use an array for looking up the mask corresponding to the 2-bit
information instead of the switch/case.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: Felix Beck <felix@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
The information from the kernel parameter is only needed during init.
Keep the three pieces (busid, wwpn and lun) local to simplify the
global zfcp_data structures. While at it, also remove the unused
loglevel variable and give the module parameter variable a better
name.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: Felix Beck <felix@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Get rid of this one:
drivers/s390/scsi/zfcp_erp.c: In function 'zfcp_erp_thread':
drivers/s390/scsi/zfcp_erp.c:1400: warning: ignoring return value of
'down_interruptible', declared with attribute warn_unused_result
zfcp_erp_thread is a kernel thread which can't receive any signals.
So introduce a dummy variable and get rid of the warning.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
After the latest changes, the list of FCP devices is only used to
lookup the adapter for requests from the actcli tool. Change this to
use the lookup function in the cio layer. Now we can remove the
adapter list and have one place less to use the global config_lock.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: Felix Beck <felix@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
When waiting for a request claim the SBAL before waiting. This way,
locking before each check of the free counter is not required and
sparse does not emit warnings for the complicated locking scheme.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: Felix Beck <felix@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Synchronize the registration and de-registration with the SCSI layer
at CCW registration, de-registration. Before we registered with the
SCSI layer on adapter activation. This way the reg and de-reg process
is in balance.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Move the closing parenthesis before the line break.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: Felix Beck <felix@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Register zfcp with the new /proc/service_level interface to report the
FCP microcode level. When the adapter goes offline or a channel path
disappears, zfcp unregisters, since the microcode version might change
and zfcp does not know about it.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The zfcp_scsi_queuecommand was not acting according to the standard
when the respective unit was not available. In this case an -EBUSY was
returned, which is not valid in itself, and in addition scsi_done
was called. This combination is not allowed and was leading to a
double finish of the request and therefor double decrement of the
host_busy counter.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Waiting for the ERP to be finished in a task running in the global
kernel work-queue is a bad idea, especially if the ERP needs to run
another job in this work-queue before it can finish. -> deadlock.
This patch removes the necessity to wait for a finished ERP from the
scan task and moves the job scheduling to the end of the ERP.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
The check of having a valid pointer was performed before the
processing was secured by the lock. Between those two steps the
pointer can turn invalid. During further processing another value is
used (referenced by the pointer described above) as a function pointer
which is never verified to be valid either, resulting under some
circumstances in an invalid function call. This patch is fixing both
issues.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Prevent a SCSI target scan for a rport which have turned invalid
in the meantime.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Aborting a SCSI cmnd might requrie to send a abort_fsf_cmnd. If the
creation of this fsf_req fails an ERR_PTR is returned where a NULL
value would be expected as an error indicator. This ERR_PTR is
dereferenced as valid fsf_req in succeeding processing leading to
an error.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Running two wka_port_get calls in parallel could issue two open_port
requests, overwriting the port handle. Don't issue an open_port
for the state PORT_OPENING, and only read the data from GOOD
responses.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
For an incoming RSCN it was checked by the ZFCP_STATUS_PORT_DID_DID
define to re-open a remote port or to test the connection. Since this
define was re-used it was also necessary to replace that define with
ZFCP_STATUS_PORT_PHYS_OPEN.
Signed-off-by: Martin Petermann <martin@linux.vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Fix multiple problems found in the hexdump data:
- length calculation was wrong, traces were incomplete
- FC payloads were dumped in different record than the output
function tried to read
- minor fixes in output
- allow complete RSCN traces (up to 1024 bytes according to spec)
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
If an open port fsf request times out (in erp) the
corresponding erp_action member of the fsf
request need to set to NULL. If the port structure
will be removed later-on there will be still a
reference in the fsf request to the non existing
erp_action otherwise.
Signed-off-by: Martin Petermann <martin.petermann@de.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Attaching a unit immediately after setting the adapter online should
be possible. The problem right now is that the port_scan runs from a
workqueue and has not finished when the set_online call returns and
the sysfs structures for the ports are not available yet. Fix that by
waiting for the port scan to complete.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Fix leftover from last typecast patch:
drivers/s390/scsi/zfcp_aux.c: In function ‘zfcp_port_enqueue’:
drivers/s390/scsi/zfcp_aux.c:629: warning: format ‘%016llx’ expects
type ‘long long unsigned int’, but argument 3 has type ‘u64’
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Fix the handling of the request list in the error path:
- Use irqsave for the lock as in the good path.
- Before removing the request, check if it is still in the list, a
call to dismiss_all might have changed the list in between.
- zfcp_qdio_send does not change the queue counters on failure,
trying revert something is wrong, so remove this.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
When allocating fsf requests without qtcb, store the pointer to the
mempool in the fsf requests for later call to mempool_free. This
codepath is only used by the status_read requests.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
The per adapter req_list_lock must be held with interrupts disabled, otherwise
we might end up with nice deadlocks as lockdep tells us (see below).
zfcp 0.0.1804: QDIO problem occurred.
=========================================================
[ INFO: possible irq lock inversion dependency detected ]
2.6.27-rc8-00035-g4a77035-dirty #86
---------------------------------------------------------
swapper/0 just changed the state of lock:
(&adapter->erp_lock){++..}, at: [<00000000002c82ae>] zfcp_erp_adapter_reopen+0x4e/0x8c
but this lock took another, hard-irq-unsafe lock in the past:
(&adapter->req_list_lock){-+..}
and interrupts could create inverse lock ordering between them.
[tons of backtraces, but only the interesting part follows]
the second lock's dependencies:
-> (&adapter->req_list_lock){-+..} ops: 2280627634176 {
initial-use at:
[<0000000000071f10>] __lock_acquire+0x504/0x18bc
[<000000000007335c>] lock_acquire+0x94/0xbc
[<00000000003d7224>] _spin_lock_irqsave+0x6c/0xb0
[<00000000002cf684>] zfcp_fsf_req_dismiss_all+0x50/0x140
[<00000000002c87ee>] zfcp_erp_adapter_strategy_generic+0x66/0x3d0
[<00000000002c9498>] zfcp_erp_thread+0x88c/0x1318
[<000000000001b0d2>] kernel_thread_starter+0x6/0xc
[<000000000001b0cc>] kernel_thread_starter+0x0/0xc
in-softirq-W at:
[<0000000000072172>] __lock_acquire+0x766/0x18bc
[<000000000007335c>] lock_acquire+0x94/0xbc
[<00000000003d7224>] _spin_lock_irqsave+0x6c/0xb0
[<00000000002ca73e>] zfcp_qdio_int_resp+0xbe/0x2ac
[<000000000027a1d6>] qdio_kick_inbound_handler+0x82/0xa0
[<000000000027daba>] tiqdio_inbound_processing+0x62/0xf8
[<0000000000047ba4>] tasklet_action+0x100/0x1f4
[<0000000000048b5a>] __do_softirq+0xae/0x154
[<0000000000021e4a>] do_softirq+0xea/0xf0
[<00000000000485de>] irq_exit+0xde/0xe8
[<0000000000268c64>] do_IRQ+0x160/0x1fc
[<00000000000261a2>] io_return+0x0/0x8
[<000000000001b8f8>] cpu_idle+0x17c/0x224
hardirq-on-W at:
[<0000000000072190>] __lock_acquire+0x784/0x18bc
[<000000000007335c>] lock_acquire+0x94/0xbc
[<00000000003d702c>] _spin_lock+0x5c/0x9c
[<00000000002caff6>] zfcp_fsf_req_send+0x3e/0x158
[<00000000002ce7fe>] zfcp_fsf_exchange_config_data+0x106/0x124
[<00000000002c8948>] zfcp_erp_adapter_strategy_generic+0x1c0/0x3d0
[<00000000002c98ea>] zfcp_erp_thread+0xcde/0x1318
[<000000000001b0d2>] kernel_thread_starter+0x6/0xc
[<000000000001b0cc>] kernel_thread_starter+0x0/0xc
}
... key at: [<0000000000e356c8>] __key.26629+0x0/0x8
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmit@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
It is possible that a remote port has a problem, the SCSI device gets
deleted after the rport timeout and then the timeout for pending SCSI
commands trigger an abort. For this case, don't delete the reference
from the SCSI device to the zfcp unit, so that we can still have the
reference to issue an abort request.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
This patch writes the channel and fabric latencies in nanoseconds per
request via blktrace for later analysis. The utilization of the inbound
and outbound adapter queue is also reported.
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Martin Peschke <mp3@de.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Some further bus_id -> dev_name() conversions in s390 code.
Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Convert most s390 users setting bus_id to dev_set_name().
css and ccw busses are deferred since they need some special
treatment.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Trace ids 107 and 3 are used twice, fix this to have unique ids for
the erp triggers.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Due to the character of a scheduled work we cannot guarantee the
LUN register to be finished before an initial device tries to use it.
Therefor we have to wait for PENDING_SCSI_WORK flag to be cleared
before proceeding.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
The zfcp_erp_thread was using the nolock version of the dbf function.
This resulted in a list access while other tasks could modifying the
list. The symptom was an erp thread running at 100% CPU and never
returning from the dbf function.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
In case of an adapter reopen all rports have to be deleted from the
environment. This should only happen for already registered rports
otherwise fc_remote_port_delete is called with a NULL pointer.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Each adapter reopen trigger automatically a scan_port task which
is waiting for the ERP to be finished before further processing.
Since the initial device setup enqueues adapter, port and LUN which
are individual ERP actions, this process would start after
everything is done. Unfortunately the port_reopen requires another
scheduled work to be finished which is queued after the automatic
scan_port -> deadlock !
This fix creates an own work queue for ERP based nameserver requests.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Now that we removed the long messages for the bit error threshold
data, put the data in the hba trace. This way, we get a short warning
for the threshold event from the hardware and have the data in the
trace for further analysis.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Reduce the size of zfcp data structures by removing unused and
redundant members. scsi_lun is only the mangled version of the
fcp_lun. So, remove the redundant field and use the fcp_lun instead.
Since the queue lock and the pci_batch indicator are only used in the
request queue, move them from the common queue struct to the adapter
struct.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Call the helper function from cio instead looping through all zfcp
adapters.
Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>