linux/drivers/md
NeilBrown ef286f6fa6 md: fix some lockdep issues between md and sysfs.
======
This fix is related to
    http://bugzilla.kernel.org/show_bug.cgi?id=15142
but does not address that exact issue.
======

sysfs does like attributes being removed while they are being accessed
(i.e. read or written) and waits for the access to complete.

As accessing some md attributes takes the same lock that is held while
removing those attributes a deadlock can occur.

This patch addresses 3 issues in md that could lead to this deadlock.

Two relate to calling flush_scheduled_work while the lock is held.
This is probably a bad idea in general and as we use schedule_work to
delete various sysfs objects it is particularly bad.

In one case flush_scheduled_work is called from md_alloc (called by
md_probe) called from do_md_run which holds the lock.  This call is
only present to ensure that ->gendisk is set.  However we can be sure
that gendisk is always set (though possibly we couldn't when that code
was originally written.  This is because do_md_run is called in three
different contexts:
  1/ from md_ioctl.  This requires that md_open has succeeded, and it
     fails if ->gendisk is not set.
  2/ from writing a sysfs attribute.  This can only happen if the
     mddev has been registered in sysfs which happens in md_alloc
     after ->gendisk has been set.
  3/ from autorun_array which is only called by autorun_devices, which
     checks for ->gendisk to be set before calling autorun_array.
So the call to md_probe in do_md_run can be removed, and the check on
->gendisk can also go.


In the other case flush_scheduled_work is being called in do_md_stop,
purportedly to wait for all md_delayed_delete calls (which delete the
component rdevs) to complete.  However there really isn't any need to
wait for them - they have already been disconnected in all important
ways.

The third issue is that raid5->stop() removes some attribute names
while the lock is held.  There is already some infrastructure in place
to delay attribute removal until after the lock is released (using
schedule_work).  So extend that infrastructure to remove the
raid5_attrs_group.

This does not address all lockdep issues related to the sysfs
"s_active" lock.  The rest can be address by splitting that lockdep
context between symlinks and non-symlinks which hopefully will happen.

Signed-off-by: NeilBrown <neilb@suse.de>
2010-02-10 11:26:09 +11:00
..
raid6test md: drivers/md/unroll.pl replaced with awk analog 2009-10-16 16:25:19 +11:00
.gitignore
bitmap.c md/bitmap: update dirty flag when bitmap bits are explicitly set. 2009-12-14 12:51:41 +11:00
bitmap.h md: Support write-intent bitmaps with externally managed metadata. 2009-12-14 12:51:41 +11:00
dm-bio-record.h dm: preserve bi_io_vec when resubmitting bios 2009-04-02 19:55:23 +01:00
dm-crypt.c dm crypt: add plain64 iv 2009-12-10 23:52:25 +00:00
dm-delay.c dm table: pass correct dev area size to device_area_is_valid 2009-07-23 20:30:42 +01:00
dm-exception-store.c dm snapshot: move cow ref from exception store to snap core 2009-12-10 23:52:12 +00:00
dm-exception-store.h dm snapshot: add merging 2009-12-10 23:52:32 +00:00
dm-io.c dm io: handle empty barriers 2009-12-10 23:52:22 +00:00
dm-ioctl.c dm: rename dm_suspended to dm_suspended_md 2009-12-10 23:52:26 +00:00
dm-kcopyd.c dm kcopyd: accept zero size jobs 2009-12-10 23:52:13 +00:00
dm-linear.c dm table: pass correct dev area size to device_area_is_valid 2009-07-23 20:30:42 +01:00
dm-log-userspace-base.c dm log: userspace fix incorrect luid cast in userspace_ctr 2009-10-16 23:18:15 +01:00
dm-log-userspace-transfer.c dm/connector: Only process connector packages from privileged processes 2009-10-02 10:54:10 -07:00
dm-log-userspace-transfer.h dm log: userspace add luid to distinguish between concurrent log instances 2009-09-04 20:40:34 +01:00
dm-log.c dm raid1: report flush errors separately in status 2009-12-10 23:52:02 +00:00
dm-mpath.c dm mpath: reject messages when device is suspended 2009-12-10 23:52:27 +00:00
dm-mpath.h dm mpath: remove is_active from struct dm_path 2008-10-10 13:36:58 +01:00
dm-path-selector.c dm: path selector use module refcount directly 2009-04-02 19:55:27 +01:00
dm-path-selector.h dm mpath: add start_io and nr_bytes to path selectors 2009-06-22 10:12:27 +01:00
dm-queue-length.c dm mpath: add queue length load balancer 2009-06-22 10:12:27 +01:00
dm-raid1.c dm raid1: explicitly initialise bio_lists 2009-12-10 23:52:06 +00:00
dm-region-hash.c dm raid1: remove bio_endio from dm_rh_mark_nosync 2009-12-10 23:52:05 +00:00
dm-round-robin.c dm mpath: add start_io and nr_bytes to path selectors 2009-06-22 10:12:27 +01:00
dm-service-time.c dm mpath: add service time load balancer 2009-06-22 10:12:28 +01:00
dm-snap-persistent.c dm exception store: add merge specific methods 2009-12-10 23:52:29 +00:00
dm-snap-transient.c dm snapshot: move cow ref from exception store to snap core 2009-12-10 23:52:12 +00:00
dm-snap.c dm snapshot: use merge origin if snapshot invalid 2009-12-10 23:52:36 +00:00
dm-stripe.c block: Optimal I/O limit wrapper 2009-09-14 08:24:52 +02:00
dm-sysfs.c dm: rename dm_suspended to dm_suspended_md 2009-12-10 23:52:26 +00:00
dm-table.c DM: Fix device mapper topology stacking 2010-01-11 14:29:20 +01:00
dm-target.c dm target: remove struct tt_internal 2009-04-02 19:55:28 +01:00
dm-uevent.c dm: avoid _hash_lock deadlock 2009-12-10 23:51:52 +00:00
dm-uevent.h dm: uevent generate events 2007-10-20 02:01:26 +01:00
dm-zero.c dm: consolidate target deregistration error handling 2009-01-06 03:04:58 +00:00
dm.c dm: export suspended state to targets 2009-12-10 23:52:27 +00:00
dm.h dm: rename dm_suspended to dm_suspended_md 2009-12-10 23:52:26 +00:00
faulty.c md: add MODULE_DESCRIPTION for all md related modules. 2009-12-14 12:51:41 +11:00
Kconfig md: revise Kconfig help for MD_MULTIPATH 2009-12-14 12:51:41 +11:00
linear.c md: add MODULE_DESCRIPTION for all md related modules. 2009-12-14 12:51:41 +11:00
linear.h md/linear: use call_rcu to free obsolete 'conf' structures. 2009-06-18 08:49:42 +10:00
Makefile md: drivers/md/unroll.pl replaced with awk analog 2009-10-16 16:25:19 +11:00
md.c md: fix some lockdep issues between md and sysfs. 2010-02-10 11:26:09 +11:00
md.h raid: improve MD/raid10 handling of correctable read errors. 2009-12-14 12:51:41 +11:00
mktables.c md/raid6: move raid6 data processing to raid6_pq.ko 2009-03-31 15:09:39 +11:00
multipath.c md: add MODULE_DESCRIPTION for all md related modules. 2009-12-14 12:51:41 +11:00
multipath.h md: remove mddev_to_conf "helper" macro 2009-06-16 16:54:21 +10:00
raid0.c md: add MODULE_DESCRIPTION for all md related modules. 2009-12-14 12:51:41 +11:00
raid0.h md: remove mddev_to_conf "helper" macro 2009-06-16 16:54:21 +10:00
raid1.c md: add MODULE_DESCRIPTION for all md related modules. 2009-12-14 12:51:41 +11:00
raid1.h md/raid1: add takeover support for raid5->raid1 2009-12-14 12:51:41 +11:00
raid5.c md: fix some lockdep issues between md and sysfs. 2010-02-10 11:26:09 +11:00
raid5.h md: fix problems with RAID6 calculations for DDF. 2009-10-16 16:27:34 +11:00
raid6algos.c md: add MODULE_DESCRIPTION for all md related modules. 2009-12-14 12:51:41 +11:00
raid6altivec.uc md: drivers/md/unroll.pl replaced with awk analog 2009-10-16 16:25:19 +11:00
raid6int.uc md: drivers/md/unroll.pl replaced with awk analog 2009-10-16 16:25:19 +11:00
raid6mmx.c md/raid6: move raid6 data processing to raid6_pq.ko 2009-03-31 15:09:39 +11:00
raid6recov.c md/raid6: move raid6 data processing to raid6_pq.ko 2009-03-31 15:09:39 +11:00
raid6sse1.c md/raid6: move raid6 data processing to raid6_pq.ko 2009-03-31 15:09:39 +11:00
raid6sse2.c md/raid6: move raid6 data processing to raid6_pq.ko 2009-03-31 15:09:39 +11:00
raid6x86.h md: fix typo in FSF address 2009-03-31 14:57:37 +11:00
raid10.c md: add MODULE_DESCRIPTION for all md related modules. 2009-12-14 12:51:41 +11:00
raid10.h md: remove mddev_to_conf "helper" macro 2009-06-16 16:54:21 +10:00
unroll.awk md: drivers/md/unroll.pl replaced with awk analog 2009-10-16 16:25:19 +11:00