Commit Graph

22 Commits

Author SHA1 Message Date
ken
41410d4d9b Silence 'out of chain frames' warnings and bump the number of frames.
mps.c:		Hide the 'out of chain frames' warning behind MPS_INFO.

mps_sas.c:	Hide the SIM queue freeze/unfreeze messages behind MPS_INFO.

mpsvar.h:	Bump the number of chain frames from 1024 to 2048.  From
		testing, it looks like this makes it less likely that we'll
		run out of chain frames, and it doesn't cost much memory
		(32K).

MFC after:	3 days
2011-02-25 17:30:25 +00:00
ken
ae423e2b0b Fix several issues with the mps(4) driver.
When the driver ran out of DMA chaining buffers, it kept the timeout for
the I/O, and I/O would stall.

The driver was not freezing the device queue on errors.

mps.c:		Pull command completion logic into a separate
		function, and call the callback/wakeup for commands
		that are never sent due to lack of chain buffers.

		Add a number of extra diagnostic sysctl variables.

		Handle pre-hardware errors for configuration I/O.
		This doesn't panic the system, but it will fail the
		configuration I/O and there is no retry mechanism.
		So the device probe will not succeed.  This should
		be a very uncommon situation, however.

mps_sas.c:	Freeze the SIM queue when we run out of chain
		buffers, and unfreeze it when more commands
		complete.

		Freeze the device queue when errors occur, so that
		CAM can insure proper command ordering.

		Report pre-hardware errors for task management
		commands.  In general, that shouldn't be possible
		because task management commands don't have S/G
		lists, and that is currently the only error path
		before we get to the hardware.

		Handle pre-hardware errors (like out of chain
		elements) for SMP requests.  That shouldn't happen
		either, since we should have enough space for two
		S/G elements in the standard request.

		For commands that end with
		MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and
		MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them
		with CAM_REQUEUE_REQ to retry them unconditionally.
		These seem to be related to back end, transport
		related problems that are hopefully transient.  We
		don't want to go through the retry count for
		something that is not a permanent error.

		Keep track of the number of outstanding I/Os.

mpsvar.h:	Track the number of free chain elements.

		Add variables for the number of outstanding I/Os,
		and I/O high water mark.

		Add variables to track the number of free chain
		buffers and the chain low water mark, as well as
		the number of chain allocation failures.

		Add I/O state flags and an attach done flag.

MFC after:	3 days
2011-02-18 17:06:06 +00:00
ken
dca3933460 In the MPS driver, during device removal processing, don't assume that
the controller firmware will return all of our commands.  Instead, keep
track of outstanding I/Os and return them to CAM once device removal
processing completes.

mpsvar.h:	Declare the new "io_list" in the mps_softc.

mps.c:		Initialize the new "io_list" in the mps softc.

mps_sas.c:	o Track SCSI I/O requests on the io_list from the
		  time of mpssas_action() through mpssas_scsiio_complete().
		o Zero out the request structures used for device
		  removal commands prior to filling them out.
		o Once the target reset task management function completes
		  during device removal processing, assume any SCSI I/O
		  commands that are still oustanding will never return
		  from the controller, and process them manually.

Submitted by:	gibbs
MFC after:	3 days
2011-02-18 17:01:57 +00:00
mdf
8045b08e4d sysctl(9) cleanup checkpoint: amd64 GENERIC builds cleanly.
Commit the rest of the devices.
2011-01-12 19:53:56 +00:00
ken
ccf27be0d1 Fix setting LUN numbers in the mps(4) driver.
Prior to this change, the addressing method wasn't getting set, and
so the LUN field could be set incorrectly in some instances.

This fix should allow for LUN numbers up to 16777215 (and return an error
for anything larger, which wouldn't fit into the flat addressing model).

Submitted by:	scottl (in part)
2010-12-11 00:36:35 +00:00
ken
2d2e9a8983 Fix an event handling bug with the mps(4) driver.
This bug manifested itself after repeated device arrivals and
departures.  The root of the problem was that the last entry in the
reply array wasn't initialized/allocated.  So every time we got
around to that event, we had a bogus address.

There were a couple more problems with the code that are also fixed:

 - The reply mechanism was being treated as sequential (indexed by
   sc->replycurindex) even though the spec says that the driver
   should use the ReplyFrameAddress field of the post queue
   descriptor to figure out where the reply is.  There is no
   guarantee that the reply descriptors will be used in sequential
   order.

 - The second word of the reply post queue descriptor wasn't being
   checked in mps_intr_locked() to make sure that it wasn't
   0xffffffff.  So the driver could potentially come across a
   partially DMAed descriptor.

 - The number of replies allocated was one less than the actual
   size of the queue.  Instead, it was the size of the number of
   replies that can be used at one time.  (Which is one less than
   the size of the queue.)

mps.c:		When initializing the entries in the reply free
		queue, make sure we initialize the full number that
		we tell the chip we have (sc->fqdepth), not the
		number that can be used at any one time (sc->num_replies).

		When allocating replies, make sure we allocate the
		number of replies that we've told the chip exist,
		not just the number that can be used simultaneously.

		Use the ReplyFrameAddress field of the post queue
		descriptor to figure out which reply is being
		referenced.  This is what the spec says to do, and
		the spec doesn't guarantee that the replies will be
		used in order.

		Put a check in to verify that the reply address passed
		back from the card is valid.  (Panic if it isn't, we'll
		panic when we try to deference the reply pointer in any
		case.)

		In mps_intr_locked(), verify that the second word of the
		post queue descriptor is not 0xffffffff in addition to
		verifying that the unused flag is not set, so we can
		make sure we didn't get a partially DMAed descriptor.

		Remove references to sc->replycurindex, it isn't needed
		now.

mpsvar.h:	Remove replycurindex from the softc, it isn't needed now.

Reviewed by:	scottl
2010-12-10 21:45:10 +00:00
kevlo
92646cacc6 Fix double ;; 2010-12-06 10:24:06 +00:00
ken
7f0ccdf947 Add Serial Management Protocol (SMP) passthrough support to CAM.
This includes support in the kernel, camcontrol(8), libcam and the mps(4)
driver for SMP passthrough.

The CAM SCSI probe code has been modified to fetch Inquiry VPD page 0x00
to determine supported pages, and will now fetch page 0x83 in addition to
page 0x80 if supported.

Add two new CAM CCBs, XPT_SMP_IO, and XPT_GDEV_ADVINFO.  The SMP CCB is
intended for SMP requests and responses.  The ADVINFO is currently used to
fetch cached VPD page 0x83 data from the transport layer, but is intended
to be extensible to fetch other types of device-specific data.

SMP-only devices are not currently represented in the CAM topology, and so
the current semantics are that the SIM will route SMP CCBs to either the
addressed device, if it contains an SMP target, or its parent, if it
contains an SMP target.  (This is noted in cam_ccb.h, since it will change
later once we have the ability to have SMP-only devices in CAM's topology.)

smp_all.c,
smp_all.h:		New helper routines for SMP.  This includes
			SMP request building routines, response parsing
			routines, error decoding routines, and structure
			definitions for a number of SMP commands.

libcam/Makefile:	Add smp_all.c to libcam, so that SMP functionality
			is available to userland applications.

camcontrol.8,
camcontrol.c:		Add smp passthrough support to camcontrol.  Several
			new subcommands are now available:

			'smpcmd' functions much like 'cmd', except that it
			allows the user to send generic SMP commands.

			'smprg' sends the SMP report general command, and
			displays the decoded output.  It will automatically
			fetch extended output if it is available.

			'smppc' sends the SMP phy control command, with any
			number of potential options.  Among other things,
			this allows the user to reset a phy on a SAS
			expander, or disable a phy on an expander.

			'smpmaninfo' sends the SMP report manufacturer
			information and displays the decoded output.

			'smpphylist' displays a list of phys on an
			expander, and the CAM devices attached to those
			phys, if any.

cam.h,
cam.c:			Add a status value for SMP errors
			(CAM_SMP_STATUS_ERROR).

			Add a missing description for CAM_SCSI_IT_NEXUS_LOST.

			Add support for SMP commands to cam_error_string().

cam_ccb.h:		Rename the CAM_DIR_RESV flag to CAM_DIR_BOTH.  SMP
			commands are by nature bi-directional, and we may
			need to support bi-directional SCSI commands later.

			Add the XPT_SMP_IO CCB.  Since SMP commands are
			bi-directional, there are pointers for both the
			request and response.

			Add a fill routine for SMP CCBs.

			Add the XPT_GDEV_ADVINFO CCB.  This is currently
			used to fetch cached page 0x83 data from the
			transport later, but is extensible to fetch many
			other types of data.

cam_periph.c:		Add support in cam_periph_mapmem() for XPT_SMP_IO
			and XPT_GDEV_ADVINFO CCBs.

cam_xpt.c:		Add support for executing XPT_SMP_IO CCBs.

cam_xpt_internal.h:	Add fields for VPD pages 0x00 and 0x83 in struct
			cam_ed.

scsi_all.c:		Add scsi_get_sas_addr(), a function that parses
			VPD page 0x83 data and pulls out a SAS address.

scsi_all.h:		Add VPD page 0x00 and 0x83 structures, and a
			prototype for scsi_get_sas_addr().

scsi_pass.c:		Add support for mapping buffers in XPT_SMP_IO and
			XPT_GDEV_ADVINFO CCBs.

scsi_xpt.c:		In the SCSI probe code, first ask the device for
			VPD page 0x00.  If any VPD pages are supported,
			that page is required to be implemented.  Based on
			the response, we may probe for the serial number
			(page 0x80) or device id (page 0x83).

			Add support for the XPT_GDEV_ADVINFO CCB.

sys/conf/files:		Add smp_all.c.

mps.c:			Add support for passing in a uio in mps_map_command(),
			so we can map a S/G list at once.

			Add support for SMP passthrough commands in
			mps_data_cb().  SMP is a special case, because the
			first buffer in the S/G list is outbound and the
			second buffer is inbound.

			Add support for warning the user if the busdma code
			comes back with more buffers than will work for the
			command.  This will, for example, help the user
			determine why an SMP command failed if busdma comes
			back with three buffers.

mps_pci.c:		Add sys/uio.h.

mps_sas.c:		Add the SAS address and the parent handle to the
			list of fields we pull from device page 0 and cache
			in struct mpssas_target.  These are needed for SMP
			passthrough.

			Add support for the XPT_SMP_IO CCB.  For now, this
			CCB is routed to the addressed device if it supports
			SMP, or to its parent if it does not and the parent
			does.  This is necessary because CAM does not
			currently support SMP-only nodes in the topology.

			Make SMP passthrough support conditional on
			__FreeBSD_version >= 900026.  This will make it
			easier to MFC this change to the driver without
			MFCing the CAM changes as well.

mps_user.c:		Un-staticize mpi_init_sge() so we can use it for
			the SMP passthrough code.

mpsvar.h:		Add a uio and iovecs into struct mps_command for
			SMP passthrough commands.

			Add a cm_max_segs field to struct mps_command so
			that we can warn the user if busdma comes back with
			too many segments.

			Clear the cm_reply when a command gets freed.  If
			it is not cleared, reply frames will eventually get
			freed into the pool multiple times and corrupt the
			pool.  (This fix is from scottl.)

			Add a prototype for mpi_init_sge().

sys/param.h:		Bump __FreeBSD_version to 900026 for the for the
			inclusion of the XPT_GDEV_ADVINFO and XPT_SMP_IO
			CAM CCBs.
2010-11-30 22:39:46 +00:00
mdf
9a2d058f58 Currently only opt_compat.h is included by the mps(4) driver. Also
enable /dev/mps0, which was missing from my previous patches enabling
f/w upload and download.

opt_compat.h issue noticed by scottl.
2010-10-15 15:24:59 +00:00
mdf
b1b265cb2e Fixes to mps_user_command():
- fix the leak of command struct on error
 - simplify the cleanup logic
 - EINPROGRESS is not a fatal error
 - buggy comment and error message

Reviewed by:   ken
2010-10-14 23:26:08 +00:00
mdf
212f93db80 Support firmware download. 2010-10-14 16:44:44 +00:00
mdf
c3e059acee Re-work the internals of adding items to the driver's scatter-gather
list.  Use the new internals to simplify adding transaction context
elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
mdf
59b6a24cb2 Always set cm_complete_data before calling mps_config_complete().
Reviewed by:	ken
2010-10-12 19:24:29 +00:00
mdf
a743f5be92 Re-work the internal user ioctl command table, and support the FW_UPLOAD
command.

Reviewed by:	ken (previous version)
2010-10-11 22:44:15 +00:00
mdf
be9043ab83 Add function prototypes for static functions.
Requested by:	ken
2010-10-11 22:44:05 +00:00
mdf
d9c65428dc Fix a memory leak and locking inconsistency in mps(4) ioctl handling.
Check copyin(9) for error and sanity check the length before copyin.

Reviewed by:	ken
2010-10-11 21:38:31 +00:00
mdf
bb8cecd58f Fix up the COMPAT_FREEBSD32 ioctl logic for mps(4).
Reviewed by:	ken
2010-10-11 21:26:24 +00:00
ken
aa3acd013b Turn on serialization of task management commands going down to the
controller, but make it optional.

After a problem report from Andrew Boyer, it looks like the LSI
chip may have issues (the watchdog timer fired) if too many aborts
are sent down to the chip at the same time.  We know that task
management commands are serialized, and although the manual doesn't
say it, it may be a good idea to just send one at a time.

But, since I'm not certain that this is necessary, add a tunable
and sysctl variable (hw.mps.%d.allow_multiple_tm_cmds) to control
the driver's behavior.

mps.c:		Add support for the sysctl and tunable, and add a
		comment about the possible return values to
		mps_map_command().

mps_sas.c:	Run all task management commands through two new
		routines, mpssas_issue_tm_request() and
		mpssas_complete_tm_request().

		This allows us to optionally serialize task
		management commands.  Also, change things so that
		the response to a task management command always
		comes back through the callback.  (Before it could
		come via the callback or the return value.)

mpsvar.h:	Add softc variables for the list of active task
		management commands, the number of active commands,
		and whether we should allow multiple active task
		management commands.  Add an active command flag.

mps.4:		Describe the new sysctl/loader tunable variable.

Sponsored by:	Spectra Logic Corporation
2010-10-07 21:56:10 +00:00
ken
6117b822d5 Fix a couple of mps problems.
When the driver is completely saturated with commands (1024 in the
case of the SAS2008 in my test system), I/O stops.  If we tell CAM
that we have one less command slot than we have actually allocated,
everything works fine.  We also need a few extra command slots to
allow for aborts and other task management commands to be sent down.

This needs more investigation to determine the root cause, but for
now this fixes things in my testing.

mps.c:		Change a printf() to mps_printf().

mps_sas.c:	Subtract 5 command slots when we tell CAM how many
		commands we can handle.

		Add some commented-out logic to print the contents
		the CDBs for timed-out commands.  This can help
		in debugging devices that are timing out.  This
		will be uncommented once I bring some CAM changes in.

Reported by:	Andrew Boyer <aboyer at averesystems dot com>
2010-09-17 21:53:56 +00:00
ken
75468fb55b MFp4 (//depot/projects/mps/...):
According to the MPT2 spec, task management commands are
serialized, and so no I/O should start while task management
commands are active.

So, to comply with that, freeze the SIM queue before we send any
task management commands (abort, target reset, etc.) down to the
IOC.  We unfreeze the queue once the task management command
completes.

It isn't clear from the spec whether multiple simultaneous task
management commands are supported.  Right now it is possible to
have multiple outstanding task management commands, especially in
the abort case.  Multiple outstanding aborts do complete
successfully, so it may be supported.

We also don't yet have any recovery mechanism (e.g. reset the IOC)
if the task management command fails.
2010-09-16 22:54:56 +00:00
ken
871d0bf87f MFp4: (//depot/projects/mps/...)
Report data overruns properly.

Submitted by:	scottl
2010-09-14 17:22:06 +00:00
ken
f65b2b217c MFp4 (//depot/projects/mps/...)
Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers.

This driver supports basic I/O, and works with SAS and SATA drives and
expanders.

Basic error recovery works (i.e. timeouts and aborts) as well.

Integrated RAID isn't supported yet, and there are some known bugs.

So this isn't ready for production use, but is certainly ready for
testing and additional development.  For the moment, new commits to this
driver should go into the FreeBSD Perforce repository first
(//depot/projects/mps/...) and then get merged into -current once
they've been vetted.

This has only been added to the amd64 GENERIC, since that is the only
architecture I have tested this driver with.

Submitted by:	scottl
Discussed with:	imp, gibbs, will
Sponsored by:	Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00