Move the mid-layer's ->queuecommand() invocation from being locked
with the host lock to being unlocked to facilitate speeding up the
critical path for drivers who don't need this lock taken anyway.
The patch below presents a simple SCSI host lock push-down as an
equivalent transformation. No locking or other behavior should change
with this patch. All existing bugs and locking orders are preserved.
Additionally, add one parameter to queuecommand,
struct Scsi_Host *
and remove one parameter from queuecommand,
void (*done)(struct scsi_cmnd *)
Scsi_Host* is a convenient pointer that most host drivers need anyway,
and 'done' is redundant to struct scsi_cmnd->scsi_done.
Minimal code disturbance was attempted with this change. Most drivers
needed only two one-line modifications for their host lock push-down.
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Acked-by: James Bottomley <James.Bottomley@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Adding function name in original debug prints and few more debug prints are
added.
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Cc: Stable Tree <stable@kernel.org>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Added sanity checks before accessing vdevice and added vdevice->deleted
setting for mptfc.
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
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>
After discussing this patch with LSI, I resubmitting with a recommended
40 second wait for the alternate ioc's initialization to complete.
--
Fusion FC chips are two function with some shared resources. During
initialization of one function its driver inhibits the ability of the
other function's driver to allocate message frames by clearing its
"active" flag. Should mid-layer error recovery be initiated for a
scsi command during this initialization (which can take up to 40 seconds)
error recovery will escalate to the level of host reset. This host
reset might fail (as the other function is resetting) resulting in
all connected targets being taken offline.
This patch holds off mid-layer error recovery for up to 40 seconds
to permit initialization of the other function to complete.
Signed-off-by: Michael Reed <mdr@sgi.com>
Acked-by: "Desai, Kashyap" <Kashyap.Desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
1) rewrite of ioctl_cmds internal generated function that issue commands to
firmware, porting them to be single threaded using the generic MPT_MGMT
struct. All wait Queues are replace by completion Queue.
2) added seperate callback handler for ioctl task managment
(mptctl_taskmgmt_reply), to handle command that timeout
3) rewrite mptctl_bus_reset
Signed-off-by: Kashyap Desai <kadesai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
1.) Added taskmgmt_quiesce_io flag in IOC and removed resetPending from
_MPT_SCSI_HOST struct.
2.) Reset from Scsi mid layer and internal Reset are seperate context.
Adding DeviceResetCtx for internal Device reset frame.
mptsas_taskmgmt_complete is optimized as part of implementation.
Signed-off-by: Kashyap Desai <kadesai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Rewrite of all internal generated functions that issue commands to firmware,
porting them to be single threaded using the generic MPT_MGMT
struct. Implemented using completion Queue.
Signed-off-by: Kashyap Desai <kadesai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
The reason for this change is there is a data corruption when four different
physical memory regions in the 36GB to 37GB region are
accessed. This is only affecting 1078.
The solution is we need to use different addressing when filling in
the scatter gather table for the effected memory regions. So instead
of snooping on all four different memory holes, we treat any physical
addresses in the 36GB address with the same algorithm.
The fix is explained below
1) Ensure that the message frames are NOT located in the trouble
region. There is no remapping available for message frames, they must
be allocated outside the problem region.
2) Ensure that Sense buffers are NOT in the trouble region. There is
no remapping available.
3) Walk through the SGE entries and if any are inside the trouble region
then they need to be remapped as discussed below.
1) Set the Local Address bit in the SGE Flags field.
MPI_SGE_FLAGS_LOCAL_ADDRESS
2) Ensure we are using 64-bit SGEs
3) Set MSb (Bit 63) of the 64-bit address, this will indicate buffer
location is Host Memory.
Signed-off-by: Kashyap Desai <kadesai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
[jejb: fixed up a ton of missed conversions.
All of you are on notice this has happened, driver trees will now
need to be rebased]
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: SCSI List <linux-scsi@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Kobjects do not have a limit in name size since a while, so stop
pretending that they do.
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Updating copyright statement to include the year 2008
Signed-off-by: Sathya Prakash <sathya.prakash@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Following a hard reset of a SAS raid, one of the raid targets is occasionally
missing. I tracked this down to a pretty obscure little bug.
The LSI fusion drivers for SAS and Fibre Channel both use their respective
transport layers. Those transport layers increment the target number
assigned to new targets.
The routine __scsi_scan_target uses the "this_id" element of the Scsi_Host
structure to avoid scanning the scsi host adapter. Both fusion drivers set
"this_id" from a value returned in a firmware PortFacts response. For my
particular test case (SAS) the firmware id assigned to the initiator was
173. After enough raid resets to cause the raid targets to go and come a
sufficient number of times, the id assigned by the transport to a raid
target would match the id assigned by the host adapter to the "this_id"
field, resulting in that target not being scanned.
Fix by not assigning this_id and not checking it in slave_configure.
Signed-off-by: Michael Reed <mdr@sgi.com>
Acked-by: "Moore, Eric" <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
ScsiLookup is an array of pending scmd pointers that the scsi lld
maintains. This array is touched from queuecommand, eh threads, and
interrupt context. This array should put under locks, hence this patch
to synchronize its access. I've added some nice little function
wrappers for this, and moved the ScsiLookup array over to MPT_ADAPTER
struct.
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
The driver is currently typecasting to obtain the shost hostdata. The
driver is updated to use the shost_priv macro.
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Cleaning up code by accesing the ioc pointer directly instead of via hd->ioc. In the future, most data members of struct MPT_SCSI_HOST will be either deleted or moved to struct MPT_ADAPTER.
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
common naming of vdevice through out driver
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Standardize all prints using common MYIOC_s_XXX_FMT macro defined in mptbase.h. Currently the driver uses several different methods to display info, where in some cases the "controller name" generating the printk is not provided.
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Recently LSI Logic Corp was renamed as LSI Corp, so whereever there is
a reference of LSI Logic, it is changed to LSI in mpt fusion driver
code.
signed-off-by: Sathya Prakash <sathya.prakash@lsi.com>
Acked-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
When there is state change in FC links, a message is displayed with
old and new link speed.
signed-off-by: Sathya Prakash <sathya.prakash@lsi.com>
Acked-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
The call back index requires only u8 but in lot of places it is
referred as int, now everywhere the call back index variables are
declared as u8 with uniform name cb_idx
signed-off-by: Sathya Prakash <sathya.prakash@lsi.com>
Acked-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
This patch contains changes in mptfc.c, mptlan.c, mptsas.c and mptspi.c to
support logging in MPT fusion drivers.
The changes are majorly in debug printks, the existing debugprintk are
modified accroding to new debug macros defined in the file mptbdebug.h
signed-off-by: Sathya Prakash <sathya.prakash@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Add support for Brocade 410/420 4Gbit FC HBAs.
They are re-branded LSI HBAs [LSI7104EP-LC/LSI7204EP-LC]
Signed-off-by: Sathya Prakash <sathya.prakash@lsi.com>
Acked-by: "Moore, Eric" <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
New sysfs scsi_host attributes are added to provide information about Firmware
version, BIOS version, MPI version and other product related information
signed-off-by: Sathya Praksh <sathya.prakash@lsi.com>
Acked-by: "Moore, Eric" <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Update assocated fusion sources with new support email address.
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
After Al Viro (finally) succeeded in removing the sched.h #include in module.h
recently, it makes sense again to remove other superfluous sched.h includes.
There are quite a lot of files which include it but don't actually need
anything defined in there. Presumably these includes were once needed for
macros that used to live in sched.h, but moved to other header files in the
course of cleaning it up.
To ease the pain, this time I did not fiddle with any header files and only
removed #includes from .c-files, which tend to cause less trouble.
Compile tested against 2.6.20-rc2 and 2.6.20-rc2-mm2 (with offsets) on alpha,
arm, i386, ia64, mips, powerpc, and x86_64 with allnoconfig, defconfig,
allmodconfig, and allyesconfig as well as a few randconfigs on x86_64 and all
configs in arch/arm/configs on arm. I also checked that no new warnings were
introduced by the patch (actually, some warnings are removed that were emitted
by unnecessarily included header files).
Signed-off-by: Tim Schmielau <tim@physik3.uni-rostock.de>
Acked-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
bump version, and fix email addr for lsi support
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Add support for greater than 255 target and luns.
Kill the hd->Target[] field, and change all references
of bus_id/target_id, to channel/id.
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
* Add modinfo driver version support.
* Change copyright year to 2007.
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Thanks to James Smart for the inspiration.
Stall error handler if attempting recovery while an rport is blocked.
This avoids device offline scenarios due to errors in the error handler.
Also verify that VirtDevice is available before issuing scsi command.
VirtDevice is removed when fc transport removes a target.
See James Smart's patch of 08/17/2006 for greater detail.
http://marc.theaimsgroup.com/?l=linux-scsi&m=115583213624803&w=2
Also bump version number per Eric's request.
Signed-off-by: Michael Reed <mdr@sgi.com>
Acked-by: Eric Moore <eric.moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Add host_supported_speeds, host_maxframe_size, host_speed, host_fabric_name,
host_port_type, host_port_state, and host_symbolic_name transport attributes
to fusion fibre channel.
Signed-off-by: Michael Reed <mdr@sgi.com>
Acked-by: Moore, Eric <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
This patch corrects a problem in mptfc which can result in targets
being removed after executing an "lsiutil 99" reset of the fibre
channel ports.
The last rescan event was being processed before the setup reset work
due to an inappropriate optimization in the event processing logic.
Every rescan event is now queued for execution and the setup reset
work now executes in the proper sequence.
Signed-off-by: Michael Reed <mdr@sgi.com>
Acked-by: Moore, Eric <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Based upon a conversation I had with LSI's fibre channel firmware guru,
this patch adds another condition under which the driver waits for the
firmware link initialization / target discovery to complete.
Signed-off-by: Michael Reed <mdr@sgi.com>
Acked-by: Moore, Eric <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Conflicts:
drivers/scsi/aacraid/comminit.c
Fixed up by removing the now renamed CONFIG_IOMMU option from
aacraid
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
* Adding 1078 ROC (Raid On Chip) Support - New host adapter
* Moving all PCI Vendor/Device ids to using internal defines; a request
from Christoph/James B. some time ago for when the next chip was added.
* Removing SAS 1066/1066E Vendor/Device IDs, as there are no plans to
manufacture that controller.
Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
The driver uses msleep_interruptible() in the code path responsible
for resetting the card's ports via the lsiutil command. If a
<ctrl-c> is received during the reset it can leave a port in such
a state that the only way to regain its use is to reboot the system.
Changing from msleep_interruptible() to msleep() corrects the problem.
Signed-off-by: Michael Reed <mdr@sgi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
While doing board reset testing I was able to put the system in
an infinite request/response loop between the scsi layer and
mptscsih_qcmd() by aborting the reset. This patch installs
a "SETUP RESET" handler which calls fc_remote_port_delete()
for all registered rports. This blocks the target which
prevents the loop. Additionally, should the reset fail to
complete, the transport will now terminate i/o to the target.
Signed-off-by: Michael Reed <mdr@sgi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
The fibre channel firmware provides a timer which is similar in purpose
to the fibre channel transport's device loss timer. The effect of this
timer is to extend the total time that a target will be missing beyond
the value associated with the transport's timer. This patch changes
the firmware timer to a default of one second which significantly reduces
the lag between when a target goes missing and the notification of the
fibre channel transport.
Signed-off-by: Michael Reed <mdr@sgi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Move fibre channel event and reset handling to mptfc. This will
result in fewer changes over time that need to be applied to
either mptbase.c or mptscsih.c.
Signed-off-by: Michael Reed <mdr@sgi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
A race condition exists in mptfc between the thread registering a device
with the fc transport and the scan work generated by the transport.
This race existed prior to the application of the mptfc bug fix patch.
mptfc_register_dev() calls fc_remote_port_add() with the FC_RPORT_ROLE_TARGET
bit set in the rport ids passed to the function. Having this bit set causes
fc_remote_port_add() to schedule a scan of the device.
This scan can execute before mptfc_register_dev() can fill in the dd_data
in the rport structure. When this happens, mptfc_target_alloc() will fail
because dd_data is null.
Attached is a patch which fixes the problem. The patch changes the rport ids
passed to fc_remote_port_add() to not have the TARGET bit set. This prevents
the scan from being scheduled. After mptfc_register_dev() fills in the rport
dd_data field, fc_remote_port_rolechg() is called, changing the role of the
rport to TARGET. Thus, the scan is scheduled after dd_data is filled
in which prevents the failure in mptfc_target_alloc().
Signed-off-by: Michael Reed <mdr@sgi.com>
Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
mptbase.h
bump version number to 3.03.09
remove unneeded flags
define workq and remove old fc specific locks
mptbase.c
initialize new lock and don't initialize two removed locks
mptscsih.c
when firmware reports target is no longer there, return
DID_REQUEUE for fc hosts so that i/o doesn't get killed until
the transport has an opportunity to manage the loss via its
dev loss timer
when the "eh_abort" routine is called, check to see if the
driver has the command or not before looking to see if a reset
is pending. James Smart and I talked about this and believe
that the API for this routine is: if driver doesn't have
command, return SUCCESS. This change helps prevent a target
from being taken offline. SUCCESS is returned because it's
likely that the command completed after error recovery timed
it out but before it could be aborted.
provide a routine to queue work to newly created workq, and
use it.
remove "ioc" from mptscsih_abort() it was only used one time.
the other references were via hd->ioc, so I just moved it....
net change in references to ioc via hd->ioc is zero
move hd->resetPending test and hd->timeouts increment to after
the test for whether the command to be aborted remains known
to the driver
Make certain that the workq exists before queuing work to it.
mptfc.c
no longer need to lock rport data structures as I was able to
single thread the code! I fixed up the debug code to
eliminate compilation messages due to type mismatch in the
printk. Got rid of some no longer needed rport flags.
Initialize and destroy the workq used for the rescan work.
simplify the logic regarding the increment of
fc_rescan_work_count. use post increment and test for zero
vs. pre increment and test for one; eliminate work_count
variable: queue_work can be called with the work_lock held as
it doesn't sleep
Signed-off-by: Michael Reed <mdr@sgi.com>
Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
It makes no sense in keeping the target_id and bus_id
in the VirtDevice structure, when it can be obtained
from the VirtTarget structure.
In addition, this patch fix's couple compilation bugs
in mptfc.c when MPT_DEBUG_FC is enabled. This
provided by Micheal Reed.
Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Jeff Garzik