This is the raid10 equivalent of
commit 4f0a5e012c
MD RAID1: Further conditionalize 'fullsync'
If a device in a newly assembled array is not fully recovered we
currently do a fully resync by don't need to.
Signed-off-by: NeilBrown <neilb@suse.de>
When adding a new device into an array it is normally important to
clear any stale data from ->recovery_offset else the new device may
not be recovered properly.
However when re-adding a device which is known to be nearly in-sync,
this is not needed and can be detrimental. The (bitmap-based)
resync will still happen, and further recovery is only needed from
where-ever it was already up to.
So if save_raid_disk is set, signifying a re-add, don't clear
->recovery_offset.
Signed-off-by: NeilBrown <neilb@suse.de>
Since commit d70ed2e4fa
MD: Allow restarting an interrupted incremental recovery.
we don't write out the metadata to devices while they are recovering.
This had a good reason, but has unfortunate consequences. This patch
changes things to make them work better.
At issue is what happens if the array is shut down while a recovery is
happening, particularly a bitmap-guided recovery.
Ideally the recovery should pick up where it left off.
However the metadata cannot represent the state "A recovery is in
process which is guided by the bitmap".
Before the above mentioned commit, we wrote metadata to the device
which said "this is being recovered and it is up to <here>". So after
a restart, a full recovery (not bitmap-guided) would happen from
where-ever it was up to.
After the commit the metadata wasn't updated so it still said "This
device is fully in sync with <this> event count". That leads to a
bitmap-based recovery following the whole bitmap, which should be a
lot less work than a full recovery from some starting point. So this
was an improvement.
However updates some metadata but not all leads to other problems.
In particular, the metadata written to the fully-up-to-date device
record that the array has all devices present (even though some are
recovering). So on restart, mdadm wants to find all devices and
expects them to have current event counts.
Obviously it doesn't (some have old event counts) so (when assembling
with --incremental) it waits indefinitely for the rest of the expected
devices.
It really is wrong to not update all the metadata together. Do that
is bound to cause confusion.
Instead, we should make it possible to record the truth in the
metadata. i.e. we need to be able to record that a device is being
recovered based on the bitmap.
We already have a Feature flag to say that recovery is happening. We
now add another one to say that it is a bitmap-based recovery.
With this we can remove the code that disables the write-out of
metadata on some devices.
So this patch:
- moves the setting of 'saved_raid_disk' from add_new_disk to
the validate_super methods. This makes sure it is always set
properly, both when adding a new device to an array, and when
assembling an array from a collection of devices.
- Adds a metadata flag MD_FEATURE_RECOVERY_BITMAP which is only
used if MD_FEATURE_RECOVERY_OFFSET is set, and record that a
bitmap-based recovery is allowed.
This is only present in v1.x metadata. v0.90 doesn't support
devices which are in the middle of recovery at all.
- Only skips writing metadata to Faulty devices.
- Also allows rdev state to be set to "-insync" via sysfs.
This can be used for external-metadata arrays. When the
'role' is set the device is assumed to be in-sync. If, after
setting the role, we set the state to "-insync", the role is
moved to saved_raid_disk which effectively says the device is
partly in-sync with that slot and needs a bitmap recovery.
Cc: Andrei Warkentin <andreiw@vmware.com>
Signed-off-by: NeilBrown <neilb@suse.de>
If an array is started degraded, and then the missing device
is found it can be re-added and a minimal bitmap-based recovery
will bring it fully up-to-date.
If the array is read-only a recovery would not be allowed.
But also if the array is read-only and the missing device was
present very recently, then there could be no need for any
recovery at all, so we simply include the device in the read-only
array without any recovery.
However... if the missing device was removed a little longer ago
it could be missing some updates, but if a bitmap is present it will
be conditionally accepted pending a bitmap-based update. We don't
currently detect this case properly and will include that old
device into the read-only array with no recovery even though it really
needs a recovery.
This patch keeps track of whether a bitmap-based-recovery is really
needed or not in the new Bitmap_sync rdev flag. If that is set,
then the device will not be added to a read-only array.
Cc: Andrei Warkentin <andreiw@vmware.com>
Fixes: d70ed2e4fa
Cc: stable@vger.kernel.org (3.2+)
Signed-off-by: NeilBrown <neilb@suse.de>
commit e875ecea26
md/raid10 record bad blocks as needed during recovery.
added code to the "cannot recover this block" path to record a bad
block rather than fail the whole recovery.
Unfortunately this new case was placed *after* r10bio was freed rather
than *before*, yet it still uses r10bio.
This is will crash with a null dereference.
So move the freeing of r10bio down where it is safe.
Cc: stable@vger.kernel.org (v3.1+)
Fixes: e875ecea26
Reported-by: Damian Nowak <spam@nowaker.net>
URL: https://bugzilla.kernel.org/show_bug.cgi?id=68181
Signed-off-by: NeilBrown <neilb@suse.de>
commit 6d183de407
md/raid5: fix newly-broken locking in get_active_stripe.
simplified a BUG_ON, but removed too much so now it sometimes fires
when it shouldn't.
When the STRIPE_EXPANDING flag is set, the stripe_head might be on a
special list while multiple stripe_heads are collected, or it might
not be on any list, even a 'free' list when the refcount is zero. As
long as STRIPE_EXPANDING is set, it will be found and added back to a
list eventually.
So both of the BUG_ONs which test for the ->lru being empty or not
need to avoid the case where STRIPE_EXPANDING is set.
The patch which broke this was marked for -stable, so this patch needs
to be applied to any branch that received 6d183de4
Fixes: 6d183de407
Cc: stable@vger.kernel.org (any release to which above was applied)
Signed-off-by: NeilBrown <neilb@suse.de>
The new iobarrier implementation in raid1 (which keeps normal writes
and resync activity separate) counts every request what is not before
the current resync point in either next_window_requests or
current_window_requests.
It flags that the request is counted by setting ->start_next_window.
allow_barrier follows this model exactly and decrements one of the
*_window_requests if and only if ->start_next_window is set.
However wait_barrier(), which increments *_window_requests uses a
slightly different test for setting -.start_next_window (which is set
from the return value of this function).
So there is a possibility of the counts getting out of sync, and this
leads to the resync hanging.
So change wait_barrier() to return a non-zero value in exactly the
same cases that it increments *_window_requests.
But was introduced in 3.13-rc1.
Reported-by: Bruno Wolff III <bruno@wolff.to>
URL: https://bugzilla.kernel.org/show_bug.cgi?id=68061
Fixes: 79ef3a8aa1
Cc: majianpeng <majianpeng@gmail.com>
Signed-off-by: NeilBrown <neilb@suse.de>
If we discover a bad block when reading we split the request and
potentially read some of it from a different device.
The code path of this has two bugs in RAID10.
1/ we get a spin_lock with _irq, but unlock without _irq!!
2/ The calculation of 'sectors_handled' is wrong, as can be clearly
seen by comparison with raid1.c
This leads to at least 2 warnings and a probable crash is a RAID10
ever had known bad blocks.
Cc: stable@vger.kernel.org (v3.1+)
Fixes: 856e08e237
Reported-by: Damian Nowak <spam@nowaker.net>
URL: https://bugzilla.kernel.org/show_bug.cgi?id=68181
Signed-off-by: NeilBrown <neilb@suse.de>
commit 5d8c71f9e5
md: raid5 crash during degradation
Fixed a crash in an overly simplistic way which could leave
R5_WriteError or R5_MadeGood set in the stripe cache for devices
for which it is no longer relevant.
When those devices are removed and spares added the flags are still
set and can cause incorrect behaviour.
commit 14a75d3e07
md/raid5: preferentially read from replacement device if possible.
Fixed the same bug if a more effective way, so we can now revert
the original commit.
Reported-and-tested-by: Alexander Lyakas <alex.bolshoy@gmail.com>
Cc: stable@vger.kernel.org (3.2+ - 3.2 will need a different fix though)
Fixes: 5d8c71f9e5
Signed-off-by: NeilBrown <neilb@suse.de>
Trivial: remove the few stray references to css_id, which itself
was removed in v3.13's 2ff2a7d03b "cgroup: kill css_id".
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Improve cache_status to emit:
<metadata block size> <#used metadata blocks>/<#total metadata blocks>
<cache block size> <#used cache blocks>/<#total cache blocks>
...
Adding the block sizes allows for easier calculation of the overall size
of both the metadata and cache devices. Adding <#total cache blocks>
provides useful context for how much of the cache is used.
Unfortunately these additions to the status will require updates to
users' scripts that monitor the cache status. But these changes help
provide more comprehensive information about the cache device and will
simplify tools that are being developed to manage dm-cache devices --
because they won't need to issue 3 operations to cobble together the
information that we can easily provide via a single status ioctl.
While updating the status documentation in cache.txt spaces were
tabify'd.
Requested-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Acked-by: Joe Thornber <ejt@redhat.com>
dm_btree_find_lowest_key is the reciprocal of dm_btree_find_highest_key.
Factor out common code for dm_btree_find_{highest,lowest}_key.
dm_btree_find_lowest_key is needed for an upcoming DM target, as such it
is best to get this interface in place.
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
We need to return -EINTR after a split because we invalidated iterators
(and freed the btree node) - but if we were finished inserting, we don't
want to redo the traversal.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
When deciding what order to reuse buckets we take into account both the bucket's
priority (which indicates lru order) and also the amount of live data in that
bucket. The way they were scaled together wasn't as correct as it could be...
this patch improves and documents it.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Checks if two keys have equivalent header fields.
(good enough for replacement or merging)
Used in bch_bkey_try_merge, and replacing a key
in the btree.
Signed-off-by: Nicholas Swenson <nks@daterainc.com>
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Added generic header checks to bch_bkey_try_merge,
which then calls the bkey specific function
Removed extraneous checks from bch_extent_merge
Signed-off-by: Nicholas Swenson <nks@daterainc.com>
We're in the process of turning bset.c into library code, so none of the code in
that file should know about struct cache_set or struct btree - so, move the
btree traversal part of the stats code to sysfs.c.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
More disentangling bset.c from the rest of the bcache code - soon, the
sorting routines won't have any dependencies on any outside structs.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Only use extent comparison for comparing extents, so we're not using
START_KEY() on other key types (i.e. btree pointers)
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Getting away from KEY_PTRS and moving toward KEY_U64s - and getting rid of magic
2s
Also - split out the part that checks against journal entry size so as to avoid
a dependancy on struct cache_set in bset.c
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Now that we've got code for raid5/6 stripe awareness, bcache just needs
to know about the stripes and when writing partial stripes is expensive
- we probably don't want to enable this optimization for raid1 or 10,
even though they have stripes. So add a flag to queue_limits.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
This error path shouldn't have been hit in practice.. and we've got reworked
reserve code coming soon so that it shouldn't _ever_ be bit... but if we've got
code for this error path it should be correct.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
We need a reserve for allocating buckets for new btree nodes - and now that
we've got multiple btrees, it really needs to be per btree.
This reworks the reserves so we've got separate freelists for each reserve
instead of watermarks, which seems to make things a bit cleaner, and it adds
some code so that btree_split() can make sure the reserve is available before it
starts.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Unnecessary since a bucket that has dirty pointers pointing to it can
never be invalidated - and skipping it is a measurable performance
boost, since the bucket gen will usually be a cache miss.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
We were unnecessarily waiting on a journal write to complete when we just needed
to start a journal write and start setting up the next one.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
The real fix is where we check the bytes we need against how much is
remaining - we also need to check for a journal entry bigger than our
buffer, we'll never write those and it would be bad if we tried to read
one.
Also improve the diagnostic messages.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
The code that handles overlapping extents that we've just read back in from disk
was depending on the behaviour of the code that handles overlapping extents as
we're inserting into a btree node in the case of an insert that forced an
existing extent to be split: on insert, if we had to split we'd also insert a
new extent to represent the top part of the old extent - and then that new
extent would get written out.
The code that read the extents back in thus not bother with splitting extents -
if it saw an extent that ovelapped in the middle of an older extent, it would
trim the old extent to only represent the bottom part, assuming that the
original insert would've inserted a new extent to represent the top part.
I still haven't figured out _how_ it can happen, but I'm now pretty convinced
(and testing has confirmed) that there's some kind of an obscure corner case
(probably involving extent merging, and multiple overwrites in different sets)
that breaks this. The fix is to change the mergesort fixup code to split extents
itself when required.
Signed-off-by: Kent Overstreet <kmo@daterainc.com>
Cc: linux-stable <stable@vger.kernel.org> # >= v3.10
When extending a metadata space map we should do the first commit whilst
still in bootstrap mode -- a mode where all blocks get allocated in the
new area.
That way the commit overhead is allocated from the newly added space.
Otherwise we risk running out of space.
With this fix, and the previous commit "dm space map common: make sure
new space is used during extend", the following device mapper testsuite
test passes:
dmtest run --suite thin-provisioning -n /resize_metadata_no_io/
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
When extending a low level space map we should update nr_blocks at
the start so the new space is used for the index entries.
Otherwise extend can fail, e.g.: sm_metadata_extend call sequence
that fails:
-> sm_ll_extend
-> dm_tm_new_block -> dm_sm_new_block -> sm_bootstrap_new_block
=> returns -ENOSPC because smm->begin == smm->ll.nr_blocks
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
There may be other parts of the kernel holding a reference on the dm
kobject. We must wait until all references are dropped before
deallocating the mapped_device structure.
The dm_kobject_release method signals that all references are dropped
via completion. But dm_kobject_release doesn't free the kobject (which
is embedded in the mapped_device structure).
This is the sequence of operations:
* when destroying a DM device, call kobject_put from dm_sysfs_exit
* wait until all users stop using the kobject, when it happens the
release method is called
* the release method signals the completion and should return without
delay
* the dm device removal code that waits on the completion continues
* the dm device removal code drops the dm_mod reference the device had
* the dm device removal code frees the mapped_device structure that
contains the kobject
Using kobject this way should avoid the module unload race that was
mentioned at the beginning of this thread:
https://lkml.org/lkml/2014/1/4/83
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
The comparison is always true and the compiler optimizes it out anyway.
Milan offered additional context relative to the original commit
784aae735d ("dm: add name and uuid to sysfs") which introduced the code:
"I think it is just relict of some experiments before I committed this
simple embedded sysfs kobj handling".
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Acked-by: Milan Broz <gmazyland@gmail.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
