Move all Fibre Channel related code to new file and cleanup the code
while doing so.
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NSwen Schillig <swen@vnet.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

drivers/s390/scsi/zfcp_dbf.c:692:2: warning: context imbalance in
'zfcp_rec_dbf_event_thread' - different lock contexts for basic block

Replace the parameter indicating if the lock is held with a new entry
function that only acquires the lock. This makes the lock handling
more visible and removes the sparse warning.
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

There is no need to pack data structures which describe the
contents of records in the new recovery trace.

lcrash currently depends on the binary format for the other traces,
removing the packed attribute from all traces would break trace
debugging with lcrash.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This change better spreads trace levels of recovery related events.
There was an overlap of traces for some recovery triggers and the
processing of recovery actions.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Store the index of the buffer in the inbound queue used to report
request completion in trace record for request coompletion.
This piece of information allows to better compare qdio and zfcp traces.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

sbal_last is more appropriate, because it matches sbal_first.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

sbal_last is confusing, as it is not the last one actually used,
but just a limit. sbal_limit is a better name.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This field is not needed, because it designates an index with a fix offset
from sbal_first. It's name is confusing anyway.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Remove some sparse warnings by telling sparse that zfcp_req_create
acquires the lock for the request queue.
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

If allocation of a status buffer failed the function incorrectly
returned 0 instead of -ENOMEM.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

When allocating memory for GID_PN nameserver requests, the allocation
function stores the pointer to the mempool, but then overwrites the
pointer via memset. Later, the wrong function to free the memory will
be called, since this is based on the stored pointer.

Fix this by first initializing the struct and then storing the pointer.
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Processing of an unsolicted status request can lead to a locking race
of the request_queue's queue_lock during the recreation of the
used up status read request while still in interrupt context
of the response handler.

Detaching the 'refill' of the long running status read requests from
the handler to a scheduled work is solving this issue.

In addition, each refill-run is trying to re-establish the full amount
of status read requests, which might have failed in earlier runs.
Signed-off-by: NSwen Schillig <swen@vnet.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

The latency information is provided on a SCSI device level (LUN)
which can be found at the following location

 /sys/class/scsi_device/<H:C:T:L>/device/cmd_latency
 /sys/class/scsi_device/<H:C:T:L>/device/read_latency
 /sys/class/scsi_device/<H:C:T:L>/device/write_latency

Each sysfs attribute provides the available data: min, max and sum for
fabric and channel latency and the number of requests processed.

An overrun of the variables is neither detected nor treated.  The file
has to be read twice to make a meaningful statement, because only the
differences of the values between the two reads can be used.  A reset
of the values can be achieved by writing to the attribute.
Signed-off-by: NSwen Schillig <swen@vnet.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Add the infrastructure to retrieve the fabric and channel latencies
from FSF commands for each SCSI command that has been processed. For
each unit, the sum, min, max and number of requests is tracked.
Signed-off-by: NSwen Schillig <swen@vnet.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

 - struct scsi_cmnd had a 16 bytes command buffer of its own.
   This is an unnecessary duplication and copy of request's
   cmd. It is probably left overs from the time that scsi_cmnd
   could function without a request attached. So clean that up.

 - Once above is done, few places, apart from scsi-ml, needed
   adjustments due to changing the data type of scsi_cmnd->cmnd.

 - Lots of drivers still use MAX_COMMAND_SIZE. So I have left
   that #define but equate it to BLK_MAX_CDB. The way I see it
   and is reflected in the patch below is.
   MAX_COMMAND_SIZE - means: The longest fixed-length (*) SCSI CDB
                      as per the SCSI standard and is not related
                      to the implementation.
   BLK_MAX_CDB.     - The allocated space at the request level

 - I have audit all ISA drivers and made sure none use ->cmnd in a DMA
   Operation. Same audit was done by Andi Kleen.

(*)fixed-length here means commands that their size can be determined
   by their opcode and the CDB does not carry a length specifier, (unlike
   the VARIABLE_LENGTH_CMD(0x7f) command). This is actually not exactly
   true and the SCSI standard also defines extended commands and
   vendor specific commands that can be bigger than 16 bytes. The kernel
   will support these using the same infrastructure used for VARLEN CDB's.
   So in effect MAX_COMMAND_SIZE means the maximum size command
   scsi-ml supports without specifying a cmd_len by ULD's
Signed-off-by: NBoaz Harrosh <bharrosh@panasas.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

The new FCP adapter statistics provide a variety of information about
the virtual adapter (subchannel). In order to collect this information
the zfcp driver is extended to query this information.

The information provided by the new FCP adapter statistics can be
fetched by reading from the following files in the sysfs filesystem

/sys/class/scsi_host/host<n>/seconds_active
/sys/class/scsi_host/host<n>/requests
/sys/class/scsi_host/host<n>/megabytes
/sys/class/scsi_host/host<n>/utilization

These are the statistics on a virtual adapter (subchannel) level.

The information provided is raw and not modified or interpreted by any
means.  No interpretation or modification of the values is done by the
zfcp driver.
Signed-off-by: NSwen Schillig <swen@vnet.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

When statistics are polled from sysfs, the statistics use the same
commands as the adapter initialization. Change the messages printed
here, so they are only printed during initialization and not for each
poll of adapter data.
Signed-off-by: NSwen Schillig <swen@vnet.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

When sending a exchange config data command, wait for a free SBAL.
This does not matter during adapter initialization, but this is
required for pulling adapter statistics during high I/O load.
Signed-off-by: NSwen Schillig <swen@vnet.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

In the case the unit is blocked, zfcp_unit_get has not been called
yet, so the error handling path should not call zfcp_unit_put.
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

The testcase
# chchp -v 0 0.da && sleep 59 && chchp -v 1 0.da
results in this deadlock situation:

STACK TRACE FOR TASK: 0x7e9a2048 (zfcperp0.0.c613)
0 schedule+816 [0x356b3c]
1 schedule_timeout+172 [0x357340]
2 wait_for_common+192 [0x3565fc]
3 flush_cpu_workqueue+116 [0x52af0]
4 flush_workqueue+116 [0x533b8]
5 fc_remote_port_add+64 [0x1c83ec]
6 zfcp_erp_thread+4534 [0x26585a]
7 kernel_thread_starter+6 [0x195d2]

STACK TRACE FOR TASK: 0x7f8ec048 (fc_wq_0)
0 schedule+816 [0x356b3c]
1 zfcp_erp_wait+104 [0x264568]
2 zfcp_scsi_slave_destroy+64 [0x261b24]
3 __scsi_remove_device+154 [0x1c24ba]
4 scsi_remove_device+62 [0x1c2512]
5 __scsi_remove_target+198 [0x1c25ea]
6 __remove_child+58 [0x1c26d6]
7 device_for_each_child+66 [0x1ab566]
8 scsi_remove_target+98 [0x1c268a]
9 run_workqueue+200 [0x5272c]
10 worker_thread+146 [0x52882]
11 kthread+140 [0x58360]
12 kernel_thread_starter+6 [0x195d2]

Remove the zfcp_erp_wait call that is not required here to prevent the
deadlock situation.
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

drivers/s390/scsi/zfcp_aux.c: In function ‘zfcp_fsf_incoming_els_rscn’:
drivers/s390/scsi/zfcp_aux.c:1379: warning: cast from pointer to integer of
different size
drivers/s390/scsi/zfcp_aux.c: In function ‘zfcp_fsf_incoming_els_plogi’:
drivers/s390/scsi/zfcp_aux.c:1432: warning: cast from pointer to integer of
different size
drivers/s390/scsi/zfcp_aux.c: In function ‘zfcp_fsf_incoming_els_logo’:
drivers/s390/scsi/zfcp_aux.c:1457: warning: cast from pointer to integer of
different size
..

Just passing pointers rids us of these warnings and improves readability.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

as seen in linux-next tree:

drivers/s390/scsi/zfcp_dbf.c: In function ‘zfcp_rec_dbf_event_thread’:
drivers/s390/scsi/zfcp_dbf.c:697: warning: passing argument 1 of ‘atomic_read’
 from incompatible pointer type

Caused by recent git commit:

commit 348447e8
Author: Martin Peschke <mp3@de.ibm.com>
Date:   Thu Mar 27 14:22:01 2008 +0100

    [SCSI] zfcp: Add trace records for recovery thread and its queues

We are not supposed to poke inside semaphore.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Acked-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

__FUNCTION__ is gcc-specific, use __func__
Signed-off-by: NHarvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>

Add missing docbook-comments for functions forming zfcp's internal
trace API.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Remove line breaks that do not conform to coding style.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Saving on line breaks, improving readability, by shortening excessive
function names and identifiers, by simplifying some functions call
chains, and by simplifying nesting of some data structure.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Unclutter the global zfcp_def.h header. Move everything required to
call into the debug feature to a new header file.
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Simplify usage of zfcp_dbf_tag() and calling functions.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Simplify usage of output function for hex dumps.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Remove obsolete output function.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Improve readability of code by using more convenient output function.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Change zfcp_dbf_timestamp() so that it just calculates timespec from
timestamp. First step to be able to rip this code out of zfcp.
Besides, this change makes it easier to rip out old-style debug view
functions.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This patch removes the now obsolete erp_dbf trace.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This patch writes trace records for various phases of a recovery action:
action being created, action being processed, action continueing
asynchronously, action gone, action timed out, action dismissed etc.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This patch allows any recovery event to be traced back to an exact
cause, e.g. a particular request identified by an id (address).
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This patch writes a trace record which provides information about state
changes for adapters, ports and units, e.g. target failure, targets becoming
online, targets being temporarily blocked due to pending recovery, targets
which have been recovered successfully etc.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This patch writes trace records which provide information about the
operation of the zfcp error recovery thread and the queues it works
on.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This patch registers the new recovery trace with the s390 debug
feature.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

Introducing helper functions that allow for code simpfifications.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This patch adds per request hardware debugging data to the trace
record which is written per request. It's a replacement for some sad
kernel message based debugging code. Considering the amount of trace
data, printk() is not suitable for this stuff. Writing binary traces
is more efficient. In addition we got all information in one place.

The QTCB trace data is only dumped for requests other than SCSI
requests.  Otherwise we would flood the trace ring buffer. We are
mostly interested in non-SCSI, recovery related requests here anyway.

This patch also works around a known hardware bug. It truncates QTCB
traces so that we do not save unused areas of the hardware trace.
Signed-off-by: NMartin Peschke <mp3@de.ibm.com>
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>