It is very likely that there are several blocks in bio, it is very
inefficient if we get their csums one by one. This patch improves
this problem by getting the csums in batch.

According to the result of the following test, the execute time of
__btrfs_lookup_bio_sums() is down by ~28%(300us -> 217us).

 # dd if=<mnt>/file of=/dev/null bs=1M count=1024
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Btrfs uses page_mkwrite to ensure stable pages during
crc calculations and mmap workloads.  We call clear_page_dirty_for_io
before we do any crcs, and this forces any application with the file
mapped to wait for the crc to finish before it is allowed to change
the file.

With compression on, the clear_page_dirty_for_io step is happening after
we've compressed the pages.  This means the applications might be
changing the pages while we are compressing them, and some of those
modifications might not hit the disk.

This commit adds the clear_page_dirty_for_io before compression starts
and makes sure to redirty the page if we have to fallback to
uncompressed IO as well.
Signed-off-by: NChris Mason <chris.mason@fusionio.com>
Reported-by: NAlexandre Oliva <oliva@gnu.org>
cc: stable@vger.kernel.org

The nodesize is capped at 64k and there are enough pages preallocated in
extent_buffer::inline_pages. The fallback to kmalloc never happened
because even on the smallest page size considered (4k) inline_pages
covered the needs.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Nobody uses these io tree ops anymore so just remove them and clean up the code
a bit.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

We'll want to merge writes so they can fill a full RAID[56] stripe, but
not necessarily reads.
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: NChris Mason <chris.mason@fusionio.com>

This is required for the device replace procedure in a later step.
Two calling functions also had to be changed to have the fs_info
pointer: repair_io_failure() and scrub_setup_recheck_block().
Signed-off-by: NStefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: NChris Mason <chris.mason@fusionio.com>

When building btrfs from kernel code, it will report:

	fs/btrfs/extent_io.h:281: warning: 'extent_buffer_page' declared inline after being called
	fs/btrfs/extent_io.h:281: warning: previous declaration of 'extent_buffer_page' was here
	fs/btrfs/extent_io.h:280: warning: 'num_extent_pages' declared inline after being called
	fs/btrfs/extent_io.h:280: warning: previous declaration of 'num_extent_pages' was here

because of the wrong declaration of inline functions.
Signed-off-by: NRobin Dong <sanbai@taobao.com>

Everytime we write out dirty pages we search for an offset in the tree,
convert the bits in the state, and then when we wait we search for the
offset again and clear the bits.  So for every dirty range in the io tree we
are doing 4 rb searches, which is suboptimal.  With this patch we are only
doing 2 searches for every cycle (modulo weird things happening).  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

There are a coule scenarios where farming metadata csumming off to an async
thread doesn't help.  The first is if our processor supports crc32c, in
which case the csumming will be fast and so the overhead of the async model
is not worth the cost.  The other case is for our tree log.  We will be
making that stuff dirty and writing it out and waiting for it immediately.
Even with software crc32c this gives me a ~15% increase in speed with O_SYNC
workloads.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

We're going to use this flag EXTENT_DEFRAG to indicate which range
belongs to defragment so that we can implement snapshow-aware defrag:

We set the EXTENT_DEFRAG flag when dirtying the extents that need
defragmented, so later on writeback thread can differentiate between
normal writeback and writeback started by defragmentation.
Original-Signed-off-by: NLi Zefan <lizf@cn.fujitsu.com>
Signed-off-by: NLiu Bo <bo.li.liu@oracle.com>

We noticed that the ordered extent completion doesn't really rely on having
a page and that it could be done independantly of ending the writeback on a
page.  This patch makes us not do the threaded endio stuff for normal
buffered writes and direct writes so we can end page writeback as soon as
possible (in irq context) and only start threads to do the ordered work when
it is actually done.  Compression needs to be reworked some to take
advantage of this as well, but atm it has to do a find_get_page in its endio
handler so it must be done in its own thread.  This makes direct writes
quite a bit faster.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>

The tree modification log needs two ways to create dummy extent buffers,
once by allocating a fresh one (to rebuild an old root) and once by
cloning an existing one (to make private rewind modifications) to it.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

A user reported a panic where we were trying to fix a bad mirror but the
mirror number we were giving was 0, which is invalid.  This is because we
don't do the transid verification until after the read, so as far as the
read code is concerned the read was a success.  So instead store the mirror
we read from so that if there is some failure post read we know which mirror
to try next and which mirror needs to be fixed if we find a good copy of the
block.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>

Since we need to read and write extent buffers in their entirety we can't use
the normal bio_readpage_error stuff since it only works on a per page basis.  So
instead make it so that if we see an io error in endio we just mark the eb as
having an IO error and then in btree_read_extent_buffer_pages we will manually
try other mirrors and then overwrite the bad mirror if we find a good copy.
This works with larger than page size blocks.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

This patch simplifies how we track our extent buffers.  Previously we could exit
writepages with only having written half of an extent buffer, which meant we had
to track the state of the pages and the state of the extent buffers differently.
Now we only read in entire extent buffers and write out entire extent buffers,
this allows us to simply set bits in our bflags to indicate the state of the eb
and we no longer have to do things like track uptodate with our iotree.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Because btrfs cow's we can end up with extent buffers that are no longer
necessary just sitting around in memory.  So instead of evicting these pages, we
could end up evicting things we actually care about.  Thus we have
free_extent_buffer_stale for use when we are freeing tree blocks.  This will
make it so that the ref for the eb being in the radix tree is dropped as soon as
possible and then is freed when the refcount hits 0 instead of waiting to be
released by releasepage.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>

We spend a lot of time looking up extent buffers from pages when we could just
store the pointer to the eb the page is associated with in page->private.  This
patch does just that, and it makes things a little simpler and reduces a bit of
CPU overhead involved with doing metadata IO.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>

A few years ago the btrfs code to support blocks lager than
the page size was disabled to fix a few corner cases in the
page cache handling.  This fixes the code to properly support
large metadata blocks again.

Since current kernels will crash early and often with larger
metadata blocks, this adds an incompat bit so that older kernels
can't mount it.

This also does away with different blocksizes for nodes and leaves.
You get a single block size for all tree blocks.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

 set_extent_bit can do exclusive locking but only when called by lock_extent*,

 Drop the exclusive bits argument except when called by lock_extent.
Signed-off-by: NJeff Mahoney <jeffm@suse.com>

 lock_extent and unlock_extent are always called with GFP_NOFS, drop the
 argument and use GFP_NOFS consistently.
Signed-off-by: NJeff Mahoney <jeffm@suse.com>

Signed-off-by: NJeff Mahoney <jeffm@suse.com>

 We encountered an issue that was easily observable on s/390 systems but
 could really happen anywhere. The timing just seemed to hit reliably
 on s/390 with limited memory.

 The gist is that when an unexpected set_page_dirty() happened, we'd
 run into the BUG() in btrfs_writepage_fixup_worker since it wasn't
 properly set up for delalloc.

 This patch does the following:
 - Performs the missing delalloc in the fixup worker
 - Allow the start hook to return -EBUSY which informs __extent_writepage
   that it should mark the page skipped and not to redirty it. This is
   required since the fixup worker can fail with -ENOSPC and the page
   will have already been redirtied. That causes an Oops in
   drop_outstanding_extents later. Retrying the fixup worker could
   lead to an infinite loop. Deferring the page redirty also saves us
   some cycles since the page would be stuck in a resubmit-redirty loop
   until the fixup worker completes. It's not harmful, just wasteful.
 - If the fixup worker fails, we mark the page and mapping as errored,
   and end the writeback, similar to what we would do had the page
   actually been submitted to writeback.
Signed-off-by: NJeff Mahoney <jeffm@suse.com>

This patch adds the possibilty to read-lock an extent even if it is already
write-locked from the same thread. btrfs_find_all_roots() needs this
capability.
Signed-off-by: NArne Jansen <sensille@gmx.net>
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

My previous patch introduced some u64 for failed_mirror variables, this one
makes it consistent again.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

write_cache_pages tries to build up a large bio to stuff down the pipe.
But if it needs to wait for a page lock, it needs to make sure and send
down any pending writes so we don't deadlock with anyone who has the
page lock and is waiting for writeback of things inside the bio.

Dave Sterba triggered this as a deadlock between the autodefrag code and
the extent write_cache_pages
Signed-off-by: NChris Mason <chris.mason@oracle.com>

While looking for a performance regression a user was complaining about, I
noticed that we had a regression with the varmail test of filebench.  This was
introduced by

0d10ee2e

which keeps us from calling writepages in writepage.  This is a correct change,
however it happens to help the varmail test because we write out in larger
chunks.  This is largly to do with how we write out dirty pages for each
transaction.  If you run filebench with

load varmail
set $dir=/mnt/btrfs-test
run 60

prior to this patch you would get ~1420 ops/second, but with the patch you get
~1200 ops/second.  This is a 16% decrease.  So since we know the range of dirty
pages we want to write out, don't write out in one page chunks, write out in
ranges.  So to do this we call filemap_fdatawrite_range() on the range of bytes.
Then we convert the DIRTY extents to NEED_WAIT extents.  When we then call
btrfs_wait_marked_extents() we only have to filemap_fdatawait_range() on that
range and clear the NEED_WAIT extents.  This doesn't get us back to our original
speeds, but I've been seeing ~1380 ops/second, which is a <5% regression as
opposed to a >15% regression.  That is acceptable given that the original commit
greatly reduces our latency to begin with.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>

If I have a range where I know a certain bit is and I want to set it to another
bit the only option I have is to call set and then clear bit, which will result
in 2 tree searches.  This is inefficient, so introduce convert_extent_bit which
will go through and set the bit I want and clear the old bit I don't want.
Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>

Add a READAHEAD extent buffer flag.
Add a function to trigger a read with this flag set.

Changes v2:
 - use extent buffer flags instead of extent state flags

Changes v5:
 - adapt to changed read_extent_buffer_pages interface
 - don't return eb from reada_tree_block_flagged if it has CORRUPT flag set
Signed-off-by: NArne Jansen <sensille@gmx.net>

read_extent_buffer_pages currently has two modes, either trigger a read
without waiting for anything, or wait for the I/O to finish. The former
also bails when it's unable to lock the page. This patch now adds an
additional parameter to allow it to block on page lock, but don't wait
for completion.

Changes v5:
 - merge the 2 wait parameters into one and define WAIT_NONE, WAIT_COMPLETE and
   WAIT_PAGE_LOCK

Change v6:
 - fix bug introduced in v5
Signed-off-by: NArne Jansen <sensille@gmx.net>

The raid-retry code in inode.c can be generalized so that it works for
metadata as well. Thus, this patch moves it to extent_io.c and makes the
raid-retry code a raid-repair code.

Repair works that way: Whenever a read error occurs and we have more
mirrors to try, note the failed mirror, and retry another. If we find a
good one, check if we did note a failure earlier and if so, do not allow
the read to complete until after the bad sector was written with the good
data we just fetched. As we have the extent locked while reading, no one
can change the data in between.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

This removes a FIXME comment and introduces the first part of nodatasum
fixup: It gets the corresponding inode for a logical address and triggers a
regular readpage for the corrupted sector.

Once we have on-the-fly error correction our error will be automatically
corrected. The correction code is expected to clear the newly introduced
EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead
of "uncorrectable" eventually.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

Currently, extent_read_full_page always assumes we are trying to read mirror
0, which generally is the best we can do. To add flexibility, pass it as a
parameter. This will be needed by scrub fixup code.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

These members are not used at all.
Signed-off-by: NXiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: NLi Zefan <lizf@cn.fujitsu.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The set/clear bit and the extent split/merge hooks only ever return 0.

 Changing them to return void simplifies the error handling cases later.

 This patch changes the hook prototypes, the single implementation of each,
 and the functions that call them to return void instead.

 Since all four of these hooks execute under a spinlock, they're necessarily
 simple.
Signed-off-by: NJeff Mahoney <jeffm@suse.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The btrfs metadata btree is the source of significant
lock contention, especially in the root node.   This
commit changes our locking to use a reader/writer
lock.

The lock is built on top of rw spinlocks, and it
extends the lock tracking to remember if we have a
read lock or a write lock when we go to blocking.  Atomics
count the number of blocking readers or writers at any
given time.

It removes all of the adaptive spinning from the old code
and uses only the spinning/blocking hints inside of btrfs
to decide when it should continue spinning.

In read heavy workloads this is dramatically faster.  In write
heavy workloads we're still faster because of less contention
on the root node lock.

We suffer slightly in dbench because we schedule more often
during write locks, but all other benchmarks so far are improved.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The extent_buffers have a very complex interface where
we use HIGHMEM for metadata and try to cache a kmap mapping
to access the memory.

The next commit adds reader/writer locks, and concurrent use
of this kmap cache would make it even more complex.

This commit drops the ability to use HIGHMEM with extent buffers,
and rips out all of the related code.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Reorder extent_buffer to remove 8 bytes of alignment padding on 64 bit
builds. This shrinks its size to 128 bytes allowing it to fit into one
fewer cache lines and allows more objects per slab in its kmem_cache.

slabinfo extent_buffer reports :-

 before:-
    Sizes (bytes)     Slabs
    ----------------------------------
    Object :     136  Total  :     123
    SlabObj:     136  Full   :     121
    SlabSiz:    4096  Partial:       0
    Loss   :       0  CpuSlab:       2
    Align  :       8  Objects:      30

 after :-
    Object :     128  Total  :       4
    SlabObj:     128  Full   :       2
    SlabSiz:    4096  Partial:       0
    Loss   :       0  CpuSlab:       2
    Align  :       8  Objects:      32
Signed-off-by: NRichard Kennedy <richard@rsk.demon.co.uk>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Remove static and global declarations and/or definitions. Reduces size
of btrfs.ko by ~3.4kB.

  text    data     bss     dec     hex filename
402081    7464     200  409745   64091 btrfs.ko.base
398620    7144     200  405964   631cc btrfs.ko.remove-all
Signed-off-by: NDavid Sterba <dsterba@suse.cz>

function prototypes without a body
Signed-off-by: NDavid Sterba <dsterba@suse.cz>

pass GFP_NOFS directly to kmem_cache_alloc
Signed-off-by: NDavid Sterba <dsterba@suse.cz>