linux-stable.git/drivers/md/md.c, branch v3.18

md: Always set RECOVERY_NEEDED when clearing RECOVERY_FROZEN

2014-11-16T22:17:46+00:00

md_check_recovery will skip any recovery and also clear
MD_RECOVERY_NEEDED if MD_RECOVERY_FROZEN is set.
So when we clear _FROZEN, we must set _NEEDED and ensure that
md_check_recovery gets run.
Otherwise we could miss out on something that is needed.

In particular, this can make it impossible to remove a
failed device from an array is the  'recovery-needed' processing
didn't happen.
Suitable for stable kernels since 3.13.

Cc: stable@vger.kernel.org (3.13+)
Reported-and-tested-by: Joe Lawrence 
Fixes: 30b8feb730f9b9b3c5de02580897da03f59b6b16
Signed-off-by: NeilBrown

md: move EXPORT_SYMBOL to after function in md.c

2014-10-14T02:08:29+00:00

Signed-off-by: NeilBrown

md: discard PRINT_RAID_DEBUG ioctl

2014-10-14T02:08:29+00:00

All the interesting information printed by this ioctl
is provided in /proc/mdstat and/or sysfs.
So it isn't needed and isn't used and would be best if it didn't
exist.

Signed-off-by: NeilBrown

md: remove MD_BUG()

2014-10-14T02:08:29+00:00

Most of the places that call this are doing so pointlessly.
A couple of the others a best replaced with WARN_ON().

Signed-off-by: NeilBrown

md: clean up 'exit' labels in md_ioctl().

2014-10-14T02:08:29+00:00

There are 4 labels and we only really need two.

Signed-off-by: NeilBrown

md: remove unnecessary test for MD_MAJOR in md_ioctl()

2014-10-14T02:08:29+00:00

unknown ioctls no longer get this deep into md_ioctl since
md_ioctl_valid() was introduced in 3.14.
So remove the test and the misleading comment.

Signed-off-by: NeilBrown

md: don't allow "-sync" to be set for device in an active array.

2014-10-14T02:08:29+00:00

If an array is active, devices can be marked 'faulty', but simply
removing the 'sync' flag is wrong.  That only makes sense
for an array which is not active (and is probably only useful
for testing anyway).

Signed-off-by: NeilBrown

md: remove unwanted white space from md.c

2014-10-14T02:08:29+00:00

My editor shows much of this is RED.

Signed-off-by: NeilBrown

md: don't start resync thread directly from md thread.

2014-10-14T02:08:28+00:00

The main 'md' thread is needed for processing writes, so if it blocks
write requests could be delayed.

Starting a new thread requires some GFP_KERNEL allocations and so can
wait for writes to complete.  This can deadlock.

So instead, ask a workqueue to start the sync thread.
There is no particular rush for this to happen, so any work queue
will do.

MD_RECOVERY_RUNNING is used to ensure only one thread is started.

Reported-by: BillStuff 
Signed-off-by: NeilBrown

md: Just use RCU when checking for overlap between arrays.

2014-10-14T02:08:28+00:00

We don't really need the full mddev_lock here, and having to
drop it is messy.
RCU is enough to protect these lists.

Signed-off-by: NeilBrown