Commit Graph

18 Commits

Author SHA1 Message Date
mav
99ff385fd0 Merge CAM locking changes from the projects/camlock branch to radically
reduce lock congestion and improve SMP scalability of the SCSI/ATA stack,
preparing the ground for the coming next GEOM direct dispatch support.

Replace big per-SIM locks with bunch of smaller ones:
 - per-LUN locks to protect device and peripheral drivers state;
 - per-target locks to protect list of LUNs on target;
 - per-bus locks to protect reference counting;
 - per-send queue locks to protect queue of CCBs to be sent;
 - per-done queue locks to protect queue of completed CCBs;
 - remaining per-SIM locks now protect only HBA driver internals.

While holding LUN lock it is allowed (while not recommended for performance
reasons) to take SIM lock.  The opposite acquisition order is forbidden.
All the other locks are leaf locks, that can be taken anywhere, but should
not be cascaded.  Many functions, such as: xpt_action(), xpt_done(),
xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM
lock to be held.

To keep compatibility and solve cases where SIM lock can't be dropped, all
xpt_async() calls in addition to xpt_done() calls are queued to completion
threads for async processing in clean environment without SIM lock held.

Instead of single CAM SWI thread, used for commands completion processing
before, use multiple (depending on number of CPUs) threads.  Load balanced
between them using "hash" of the device B:T:L address.

HBA drivers that can drop SIM lock during completion processing and have
sufficient number of completion threads to efficiently scale to multiple
CPUs can use new function xpt_done_direct() to avoid extra context switch.
Make ahci(4) driver to use this mechanism depending on hardware setup.

Sponsored by:	iXsystems, Inc.
MFC after:	2 months
2013-10-21 12:00:26 +00:00
mav
40dc6506b8 MFprojects/camlock:
r249017:
Some cosmetic things:
 - Unify device to target insertion inside xpt_alloc_device() instead of
duplicating it three times.
 - Remove extra checks for empty lists of devices and targets on release
since zero refcount check also implies it.
 - Reformat code to reduce indentation.

r249103:
 - Add lock assertions to every point where reference counters are modified.
 - When reference counters are reaching zero, add assertions that there are
no children items left.
 - Add a bit more locking to the xptpdperiphtraverse().
2013-04-04 20:31:40 +00:00
mav
368b37ab92 On SIM destruction free associated CCBs, preallocated inside xpt_get_ccb().
Before this change they were just leaked.  Fortunately USB sticks now use
only one CCB, and so leak was only 2KB per detach, while other bigger SIMs
with much more allocated CCBs are rarely detached.

MFC after:	2 weeks
2013-03-27 18:55:01 +00:00
ed
e97eae1577 Mark MALLOC_DEFINEs static that have no corresponding MALLOC_DECLAREs.
This means that their use is restricted to a single C file.
2011-11-07 06:44:47 +00:00
mav
72062fdcec MFp4: Large set of CAM inprovements.
- Unify bus reset/probe sequence. Whenever bus attached at boot or later,
CAM will automatically reset and scan it. It allows to remove duplicate
code from many drivers.
- Any bus, attached before CAM completed it's boot-time initialization,
will equally join to the process, delaying boot if needed.
- New kern.cam.boot_delay loader tunable should help controllers that
are still unable to register their buses in time (such as slow USB/
PCCard/ CardBus devices), by adding one more event to wait on boot.
- To allow synchronization between different CAM levels, concept of
requests priorities was extended. Priorities now split between several
"run levels". Device can be freezed at specified level, allowing higher
priority requests to pass. For example, no payload requests allowed,
until PMP driver enable port. ATA XPT negotiate transfer parameters,
periph driver configure caching and so on.
- Frozen requests are no more counted by request allocation scheduler.
It fixes deadlocks, when frozen low priority payload requests occupying
slots, required by higher levels to manage theit execution.
- Two last changes were holding proper ATA reinitialization and error
recovery implementation. Now it is done: SATA controllers and Port
Multipliers now implement automatic hot-plug and should correctly
recover from timeouts and bus resets.
- Improve SCSI error recovery for devices on buses without automatic sense
reporting, such as ATAPI or USB. For example, it allows CAM to wait, while
CD drive loads disk, instead of immediately return error status.
- Decapitalize diagnostic messages and make them more readable and sensible.
- Teach PMP driver to limit maximum speed on fan-out ports.
- Make boot wait for PMP scan completes, and make rescan more reliable.
- Fix pass driver, to return CCB to user level in case of error.
- Increase number of retries in cd driver, as device may return several UAs.
2010-01-28 08:41:30 +00:00
trasz
2cd9ec8ad6 Fix cam_sim_free() wakeup condition and add mtx_asserts.
Submitted by:	Christoph Mallon
Reviewed by:	scottl
Approved by:	rwatson (mentor)
Sponsored by:	FreeBSD Foundation
2008-12-19 14:33:29 +00:00
trasz
1aa4ea9cfb Add SIM refcounting. This is slightly different from what DragonFly
does - in DragonFly, it's cam_sim_release() what actually frees the
SIM; cam_sim_free does nothing more than calling cam_sim_release().
Here, we drain in cam_sim_free, waiting for refcount to drop to zero.
We cannot do the same think DragonFly does, because after cam_sim_free
returns, client would destroy the sim->mtx, and CAM would trip over
an initialized mutex.

Reviewed by:	scottl
Approved by:	rwatson (mentor)
Sponsored by:	FreeBSD Foundation
2008-12-16 16:57:33 +00:00
scottl
59a37d31ba Split the camisr into per-SIM done queues. This optimizes the locking a
little bit and allows for direct dispatch of the doneq from certain
contexts that would otherwise face recursive locking problems.
2007-04-19 14:28:43 +00:00
scottl
7605c12c0f Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will
use to synchornize and protect all data objects that are used for that
SIM.  Drivers that are not yet MPSAFE register Giant and operate as
usual.  RIght now, no drivers are MPSAFE, though a few will be changed
in the coming week as this work settles down.

The driver API has changed, so all CAM drivers will need to be recompiled.
The userland API has not changed, so tools like camcontrol do not need to
be recompiled.
2007-04-15 08:49:19 +00:00
avatar
cd4bfad3f4 - Providing fine-grained malloc statistic by replacing M_DEVBUF with
module-specific malloc types.  These should help us to pinpoint the
  possible memory leakage in the future.
- Implementing xpt_alloc_ccb_nowait() and replacing all malloc/free based
  CCB memory management with xpt_alloc_ccb[_nowait]/xpt_free_ccb.  Hopefully
  this would be helpful if someday we move the CCB allocator to use UMA
  instead of malloc().

Encouraged by:	jeffr, rwatson
Reviewed by:	gibbs, scottl
Approved by:	re (scottl)
2005-07-01 15:21:30 +00:00
imp
362fcfc1e2 Start each of the license/copyright comments with /*- 2005-01-05 22:34:37 +00:00
obrien
37ee58d305 Use __FBSDID(). 2003-06-10 17:50:20 +00:00
imp
cf874b345d Back out M_* changes, per decision of the TRB.
Approved by: trb
2003-02-19 05:47:46 +00:00
alfred
bf8e8a6e8f Remove M_TRYWAIT/M_WAITOK/M_WAIT. Callers should use 0.
Merge M_NOWAIT/M_DONTWAIT into a single flag M_NOWAIT.
2003-01-21 08:56:16 +00:00
jhb
c7124acea7 Const'ify cam_sim.sim_name to quiet warnings.
Reviewed by:	-scsi
2001-01-24 01:46:18 +00:00
peter
3b842d34e8 $Id$ -> $FreeBSD$ 1999-08-28 01:08:13 +00:00
ken
2bb789d7a3 Add a number of interrelated CAM feature enhancements and bug fixes.
NOTE:  These changes will require recompilation of any userland
applications, like cdrecord, xmcd, etc., that use the CAM passthrough
interface.  A make world is recommended.

camcontrol.[c8]:
 - We now support two new commands, "tags" and "negotiate".

	- The tags commands allows users to view the number of tagged
	  openings for a device as well as a number of other related
	  parameters, and it allows users to set tagged openings for
	  a device.

	- The negotiate command allows users to enable and disable
	  disconnection and tagged queueing, set sync rates, offsets
	  and bus width.  Note that not all of those features are
	  available for all controllers.  Only the adv, ahc, and ncr
	  drivers fully support all of the features at this point.
	  Some cards do not allow the setting of sync rates, offsets and
	  the like, and some of the drivers don't have any facilities to
	  do so.  Some drivers, like the adw driver, only support enabling
	  or disabling sync negotiation, but do not support setting sync
	  rates.

 - new description in the camcontrol man page of how to format a disk
 - cleanup of the camcontrol inquiry command
 - add support in the 'devlist' command for skipping unconfigured devices if
   -v was not specified on the command line.
 - make use of the new base_transfer_speed in the path inquiry CCB.
 - fix CCB bzero cases

cam_xpt.c, cam_sim.[ch], cam_ccb.h:

 - new flags on many CCB function codes to designate whether they're
   non-immediate, use a user-supplied CCB, and can only be passed from
   userland programs via the xpt device.  Use these flags in the transport
   layer and pass driver to categorize CCBs.

 - new flag in the transport layer device matching code for device nodes
   that indicates whether a device is unconfigured

 - bump the CAM version from 0x10 to 0x11

 - Change the CAM ioctls to use the version as their group code, so we can
   force users to recompile code even when the CCB size doesn't change.

 - add + fill in a new value in the path inquiry CCB, base_transfer_speed.
   Remove a corresponding field from the cam_sim structure, and add code to
   every SIM to set this field to the proper value.

 - Fix the set transfer settings code in the transport layer.

scsi_cd.c:

 - make some variables volatile instead of just casting them in various
   places
 - fix a race condition in the changer code
 - attach unless we get a "logical unit not supported" error.  This should
   fix all of the cases where people have devices that return weird errors
   when they don't have media in the drive.

scsi_da.c:

 - attach unless we get a "logical unit not supported" error

scsi_pass.c:

 - for immediate CCBs, just malloc a CCB to send the user request in.  This
   gets rid of the 'held' count problem in camcontrol tags.

scsi_pass.h:

 - change the CAM ioctls to use the CAM version as their group code.

adv driver:

 - Allow changing the sync rate and offset separately.

adw driver

 - Allow changing the sync rate and offset separately.

aha driver:

 - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs.

ahc driver:

 - Allow setting offset and sync rate separately

bt driver:

 - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs.

NCR driver:

 - Fix the ultra/ultra 2 negotiation bug
 - allow setting both the sync rate and offset separately

Other HBA drivers:
 - Put code in to set the base_transfer_speed field for
   XPT_GET_TRAN_SETTINGS CCBs.

Reviewed by:	gibbs, mjacob (isp), imp (aha)
1999-05-06 20:16:39 +00:00
gibbs
855593c295 CAM Transport Layer (XPT).
Submitted by:	The CAM Team
1998-09-15 06:33:23 +00:00