Commit graph

42 commits

Author SHA1 Message Date
Joseph Handzik
1304953700 cciss: Adds simple mode functionality
Signed-off-by: Joseph Handzik <joseph.t.handzik@beardog.cce.hp.com>
Acked-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-08-08 11:40:15 +02:00
Stephen M. Cameron
07d0c38e7d cciss: do not attempt to read from a write-only register
Most smartarrays will tolerate it, but some new ones don't.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>

Note: this is a regression caused by commit 1ddd5049
Cc: stable@kernel.org
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-07-09 09:04:12 +02:00
Stephen M. Cameron
3e28601fdf cciss: increase timeouts for post-reset no-ops
Just to reduce the messages about timeouts that appear.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-05-06 08:23:54 -06:00
Stephen M. Cameron
19adbb9254 cciss: increase time to wait for board reset to start
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-05-06 08:23:51 -06:00
Stephen M. Cameron
9bd3c20487 cciss: add readl after writel in interrupt mask setting code
This is to ensure the board interrupts are really off when
these functions return.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-05-06 08:23:45 -06:00
Bud Brown
1ddd504954 cciss: fix lost command issue
Under certain workloads a command may seem to get lost. IOW, the Smart Array
thinks all commands have been completed but we still have commands in our
completion queue. This may lead to system instability, filesystems going
read-only, or even panics depending on the affected filesystem. We add an
extra read to force the write to complete.

Testing shows this extra read avoids the problem.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: stable@kernel.org
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-03-23 20:47:11 +01:00
Linus Torvalds
7b0cb1bdac Merge branch 'for-2.6.38/drivers' of git://git.kernel.dk/linux-2.6-block
* 'for-2.6.38/drivers' of git://git.kernel.dk/linux-2.6-block:
  cciss: reinstate proper FIFO order of command queue list
  floppy: replace NO_GEOM macro with a function
2011-01-13 10:50:24 -08:00
Jens Axboe
e6e1ee936d cciss: reinstate proper FIFO order of command queue list
Commit 8a3173de inadvertently changed the ordering when
switching to hlists. Change to regular list heads so we
can use tail list adds, this improves performance.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-01-10 21:50:33 +01:00
Stephen M. Cameron
afa842fa64 cciss: fix board status waiting code
After a reset, we should first wait for the board to become "not ready",
and then wait for it to become "ready", instead of immediately
waiting for it to become "ready", and do this waiting *after*
restoring PCI config space registers.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-10-23 18:45:06 +02:00
Stephen M. Cameron
f70dba8366 cciss: use consistent variable names
cciss: use consistent variable names
"h", for the hba structure and "c" for the command structures.
and get rid of trivial CCISS_LOCK macro.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-08-07 18:52:30 +02:00
Stephen M. Cameron
83123cb11b cciss: factor out cciss_reset_devices()
cciss: factor out cciss_reset_devices()

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-08-07 18:52:12 +02:00
Stephen M. Cameron
e99ba13627 cciss: factor out cciss_wait_for_board_ready()
cciss: factor out cciss_wait_for_board_ready()

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-08-07 18:52:11 +02:00
Mike Miller
5e216153c3 cciss: add performant mode support for Stars/Sirius
Add a mode of controller operation called Performant Mode.  Even though
cciss has been deprecated in favor of hpsa there are new controllers due
out next year that HP must support in older vendor distros.  Vendors
require all fixes/features be upstream.  These new controllers support
only 16 commands in simple mode but support up to 1024 in performant mode.
This requires us to add this support at this late date.

The performant mode transport minimizes host PCI accesses by performinf
many completions per read.  PCI writes are posted so the host can write
then immediately get off the bus not waiting for the writwe to complete to
the target.  In the context of performant mode the host read out to a
controller pulls all posted writes into host memory ensuring the reply
queue is coherent.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-08-07 18:12:51 +02:00
Mike Miller
1d1414419f cciss: make interrupt access methods return type bool
Change the return type of our interrupt access routines to bool from
unsigned long.  It makes more sense that way.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-08-07 18:12:51 +02:00
Stephen M. Cameron
aad9fb6f2c cciss: do not use void pointer for scsi hba data
cciss: do not use void pointer for scsi hba data
and get rid of related unnecessary type casting
and delete some superfluous and misleading comments nearby.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2010-02-28 19:42:32 +01:00
Stephen M. Cameron
dccc9b563e cciss: simplify scatter gather code
cciss: simplify scatter gather code.
Instead of allocating an array of pointers to a structure
containing an SGDescriptor structure, and two other elements
that aren't really used, just allocate SGDescriptor structs.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2010-02-28 19:42:31 +01:00
dann frazier
b028461d66 cciss: remove C99-style comments
Some cleanup before the header file split-out so we don't propagate this style
into new files.

Acked-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: dann frazier <dannf@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2010-02-22 13:44:45 +01:00
Alex Chiang
32a87c0114 cciss: change Cmd_sg_list.sg_chain_dma type to dma_addr_t
A recent commit broke the ia64 build:

	Author: Don Brace <brace@beardog.cce.hp.com>
	Date:   Thu Nov 12 12:50:01 2009 -0600

	cciss: Add enhanced scatter-gather support.

because of this hunk:

	--- a/drivers/block/cciss.h
	+++ b/drivers/block/cciss.h
	+struct Cmd_sg_list {
	+       SGDescriptor_struct     *sgchain;
	+       dma64_addr_t            sg_chain_dma;
	+       int                     chain_block_size;
	+};

The issue is that dma64_addr_t isn't #define'd on ia64.

The way that we're using Cmd_sg_list.sg_chain_dma is to hold an
address returned from pci_map_single().

	+               temp64.val = pci_map_single(h->pdev,
	+                                 h->cmd_sg_list[c->cmdindex]->sgchain,
	+                                 len, dir);
	+
	+               h->cmd_sg_list[c->cmdindex]->sg_chain_dma = temp64.val;

pci_map_single() returns a dma_addr_t too.

This code will still work even on a 32-bit x86 build, where
dma_addr_t is defined to be a u32 because it will simply be
promoted to the __u64 that temp64.val is defined as.

Thus, declaring Cmd_sg_list.sg_chain_dma as dma_addr_t is safe.

Cc: Don Brace <brace@beardog.cce.hp.com>
Cc: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-11-23 09:35:06 +01:00
Don Brace
5c07a311a8 cciss: Add enhanced scatter-gather support.
cciss: Add enhanced scatter-gather support.  For controllers which
supported, more than 512 scatter-gather elements per command may
be used, and the max transfer size can be increased to 8192 blocks.

Signed-off-by: Don Brace <brace@beardog.cce.hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-11-13 08:45:54 +01:00
Stephen M. Cameron
9cef0d2f4f cciss: Dynamically allocate the drive_info_struct for each logical drive.
cciss: Dynamically allocate the drive_info_struct for each logical drive.
This reduces the size of the per-hba ctlr_info structure from 106936
bytes to 8132 bytes.  That's on 32-bit systems.  On 64-bit systems, the
improvement is even bigger.  Without this, the ctlr_info struct is so big
that the driver won't even load on a 64 bit system if CISS_MAX_LUN was
at it's current setting of 1024 logical drives.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-10-01 21:15:45 +02:00
Stephen M. Cameron
39ccf9a645 cciss: Preserve all 8 bytes of LUN ID for logical drives.
Preserve all 8 bytes of the LunID field returned
by CCISS_REPORT_LOGICAL instead of only saving 4 bytes.
This fixes a bug with logical volume addressing encountered on
an MSA2012.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-10-01 21:15:43 +02:00
Stephen M. Cameron
617e134422 cciss: Dynamically allocate struct device for each logical drive as needed.
Dynamically allocate struct device for each logical drive as needed
instead of allocating the maximum we would ever need at driver init time.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-10-01 21:15:41 +02:00
Andrew Patterson
b368c9dd65 cciss: Use one scan thread per controller and fix hang during rmmod
Replace the use of one scan kthread per controller with one per driver.
Use a queue to hold a list of controllers that need to be rescanned with
routines to add and remove controllers from the queue.

Fix locking and completion handling to prevent a hang during rmmod.

Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Acked-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-10-01 21:15:41 +02:00
scameron@beardog.cca.cpqcorp.net
72f9f1324f cciss: Remove no longer needed sendcmd reject processing code
Now that the cciss SCSI error handling routines operate with interrupts
enabled, we no longer need to maintain the list of command completions that
sendcmd() might inadvertantly scoop up, since now it only runs at driver init
time, and there won't be any other commands for it to scoop up.  So we
can remove that list and the code that adds to it and processes it.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cca.cpqcorp.net>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-06-09 05:47:43 +02:00
scameron@beardog.cca.cpqcorp.net
789a424ad1 cciss: separate error processing and command retrying code in sendcmd_withirq_core()
Separate the error processing from sendcmd_withirq_core from the code
which retries commands.  The rationale for this is that the SCSI error
handling code can then be made to use sendcmd_withirq_core, but avoid
retrying commands.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cca.cpqcorp.net>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-06-09 05:47:43 +02:00
Andrew Patterson
7fe063268e cciss: add cciss driver sysfs entries
Add sysfs entries to the cciss driver needed for the dm/multipath tools.

A file for vendor, model, rev, and unique_id is added for each logical
drive under directory /sys/bus/pci/devices/<dev>/ccissX/cXdY.  Where X =
the controller (or host) number and Y is the logical drive number.

A link from /sys/bus/pci/devices/<dev>/ccissX/cXdY/block:cciss!cXdY to
/sys/block/cciss!cXdY/device is also created.  A bus is created in
/sys/bus/cciss.  A link is created from the pci ccissX entry to
/sys/bus/cciss/devices/ccissX.  Please consider this for inclusion.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Stephen M. Cameron <scameron@beardog.cca.cpqcorp.net>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-06-02 14:48:39 +02:00
Mike Miller
0a9279cc7c cciss: kernel scan thread for MSA2012
The MSA2012 cannot inform the driver of configuration changes since all
management is out of band.  This is a departure from any storage we have
supported in the past.  We need some way to detect changes on the topology
so we implement this kernel thread.  In some instances there's nothing we
can do from the driver (like LUN failure) so just print out a message.  In
the case where logical volumes are added or deleted we call
rebuild_lun_table to refresh the driver's view of the world.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-04-07 08:12:38 +02:00
Jens Axboe
8a3173de4a cciss: switch to using hlist for command list management
This both cleans up the code and also helps detect the spurious case
of a command attempted being removed from a queue it doesn't belong
to.

Acked-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2008-12-29 08:28:43 +01:00
Mike Miller
a72da29b6c cciss: make rebuild_lun_table behave better
This patch makes the rebuild_lun_table smart enough to not rip a logical
volume out from under the OS. Without this fix if a customer is running
hpacucli to monitor their storage the driver will blindly remove and re-add
the disks whenever the utility calls the CCISS_REGNEWD ioctl. Unfortunately,
both hpacucli and ACUXE call the ioctl repeatedly. Customers have reported
IO coming to a standstill. Calling the ioctl is the problem, this patch is
the fix.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cca.cpqcorp.net>
Signed-off-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2008-08-06 12:30:03 +02:00
Mike Miller
799202cbd0 [PATCH] cciss: add support for 1024 logical volumes
Add the support for a large number of logical volumes.  We will soon have
hardware that support up to 1024 logical volumes.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-07 08:39:30 -08:00
Mike Miller
92c4231aef [PATCH] cciss: set sector_size to 2048 for performance
Change the blk_queue_max_sectors from 512 to 2048.  This helps increase
performance.

[akpm@osdl.org: s/sector_size/max_sectors/]
Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-07 08:39:29 -08:00
Mike Miller
f880632f96 [PATCH] cciss: increase number of commands on controller
Remove #define NR_CMDS and replace it w/hba[i]->nr_cmds.  Most Smart Array
controllers can support up to 1024 commands but the E200 family can only
support 128.  To prevent annoying "fifo full" messages we define nr_cmds on a
per controller basis by adding it the product table.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Acked-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-07 08:39:29 -08:00
Mike Miller (OS Dev)
00988a3514 [PATCH] cciss: support for >2TB logical volumes
Add support for logical volumes >2TB.  All SAS/SATA controllers support
large volumes.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <axboe@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-01 00:39:20 -07:00
Bjorn Helgaas
4e570309e8 [PATCH] CCISS: request all PCI resources
We should call pci_request_regions() to claim all resources the device
decodes.  Previously, we claimed only the I/O port range.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Acked-by: Jeff Garzik <jeff@garzik.org>
Acked-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 10:01:21 -07:00
Christoph Hellwig
564de74a7e [PATCH] cciss: avoid defining useless MAJOR_NR macro
This sneaked in with one of the updates.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 20:14:09 -08:00
Mike Miller
fb86a35b9d [PATCH] cciss: adds MSI and MSI-X support
This creates a new function, cciss_interrupt_mode called from
cciss_pci_init.  This function determines what type of interrupt vector to
use, i.e., MSI, MSI-X, or IO-APIC.

One noticeable difference is changing the interrupt field of the controller
struct to an array of 4 unsigned ints.  The Smart Array HW is capable of
generating 4 distinct interrupts depending on the transport method in use
during operation.  These are:

#define DOORBELL_INT 0
Used to notify the contoller of configuration updates. We only use
this feature when in polling mode.

#define PERF_MODE_INT 0
Used when the controller is in Performant Mode.

#define SIMPLE_MODE_INT 2
Used when the controller is in Simple Mode (current Linux implementation).

#define MEMQ_INT_MODE 3
Not used.

When using IO-APIC interrupts these 4 lines are OR'ed together so when any
one fires an interrupt an is generated.  In MSI or MSI-X mode this hardware
OR'ing is ignored.  We must register for our interrupt depending on what
mode the controller is running.  For Linux we use SIMPLE_MODE_INT
exclusively at this time.  Please consider this for inclusion.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <axboe@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 20:14:00 -08:00
mike.miller@hp.com
3da8b713da [SCSI] cciss: scsi error handling
This patch adds SCSI error handling code to the SCSI portion
of the cciss driver.

Signed-off-by: Stephen M. Cameron <steve.cameron@hp.com>
Acked-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-11-06 14:12:49 -06:00
Mike Miller
33079b2197 [PATCH] cciss: direct lookup for command completions
This patch changes the way we complete commands.  In the old method when we
got a completion we searched our command list from the top until we find it.

This method uses a tag associated with each command (not SCSI command tagging)
to index us directly to the completed command.  This helps performance.

Signed-off-by: Don Brace <dab@hp.com>
Signed-off-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-13 08:22:30 -07:00
Mike Miller
ddd474420a [PATCH] cciss: new disk register/deregister routines
This patch removes a couple of functions dealing with configuration and
replaces them with new functions.  This implementation fixes some bugs
associated with the ACUXE.  It also allows a logical volume to be removed from
the middle without deleting all volumes behind it.

If a user has 5 logical volumes and decides he wants to reconfigure volume
number 3, he can now do that without removing volumes 4 & 5 first.  This code
has been tested in our labs against all application software.

Signed-off-by: Chase Maupin <chase.maupin@hp.com>
Signed-off-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-13 08:22:30 -07:00
Mike Miller
1f8ef3806c [PATCH] cciss: busy_initializing flag
This patch adds a flag called busy_initializing.  If there are multiple
controllers in a server AND the HP agents are running it's possible the agents
may try to poll a card that is still initializing if the driver is removed and
then added again.

Signed-off-by: Don Brace <dab@hp.com>
Signed-off-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-13 08:22:29 -07:00
Mike Miller
ad2b93123d [PATCH] cciss per disk queue
This patch adds per disk queue functionality to cciss.  Sometime back I
submitted a patch but it looks like only part of what I needed.  In the 2.6
kernel if we have more than one logical volume the driver will Oops during
rmmod.  It seems all of the queues actually point back to the same queue.
So after deleting the first volume you hit a null pointer on the second
one.

This has been tested in our labs.  There is no difference in performance,
it just fixes the Oops.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <axboe@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-28 08:39:01 -07:00
Linus Torvalds
1da177e4c3 Linux-2.6.12-rc2
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.

Let it rip!
2005-04-16 15:20:36 -07:00