openwrt/staging/blogic.git
16 years agoBtrfs: Change find_extent_buffer to use TestSetPageLocked
Chris Mason [Wed, 25 Jun 2008 20:01:31 +0000 (16:01 -0400)]
Btrfs: Change find_extent_buffer to use TestSetPageLocked

This makes it possible for callers to check for extent_buffers in cache
without deadlocking against any btree locks held.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add btree locking to the tree defragmentation code
Chris Mason [Wed, 25 Jun 2008 20:01:31 +0000 (16:01 -0400)]
Btrfs: Add btree locking to the tree defragmentation code

The online btree defragger is simplified and rewritten to use
standard btree searches instead of a walk up / down mechanism.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Replace the transaction work queue with kthreads
Chris Mason [Wed, 25 Jun 2008 20:01:31 +0000 (16:01 -0400)]
Btrfs: Replace the transaction work queue with kthreads

This creates one kthread for commits and one kthread for
deleting old snapshots.  All the work queues are removed.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoAdd btrfs_end_transaction_throttle to force writers to wait for pending commits
Chris Mason [Wed, 25 Jun 2008 20:01:31 +0000 (16:01 -0400)]
Add btrfs_end_transaction_throttle to force writers to wait for pending commits

The existing throttle mechanism was often not sufficient to prevent
new writers from coming in and making a given transaction run forever.
This adds an explicit wait at the end of most operations so they will
allow the current transaction to close.

There is no wait inside file_write, inode updates, or cow filling, all which
have different deadlock possibilities.

This is a temporary measure until better asynchronous commit support is
added.  This code leads to stalls as it waits for data=ordered
writeback, and it really needs to be fixed.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fix snapshot deletion to release the alloc_mutex much more often.
Chris Mason [Wed, 25 Jun 2008 20:01:30 +0000 (16:01 -0400)]
Btrfs: Fix snapshot deletion to release the alloc_mutex much more often.

This lowers the impact of snapshot deletion on the rest of the FS.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add a skip_locking parameter to struct path, and make various funcs honor it
Chris Mason [Wed, 25 Jun 2008 20:01:30 +0000 (16:01 -0400)]
Btrfs: Add a skip_locking parameter to struct path, and make various funcs honor it

Allocations may need to read in block groups from the extent allocation tree,
which will require a tree search and take locks on the extent allocation
tree.  But, those locks might already be held in other places, leading
to deadlocks.

Since the alloc_mutex serializes everything right now, it is safe to
skip the btree locking while caching block groups.  A better fix will be
to either create a recursive lock or find a way to back off existing
locks while caching block groups.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoFix btrfs_next_leaf to check for new items after dropping locks
Chris Mason [Wed, 25 Jun 2008 20:01:30 +0000 (16:01 -0400)]
Fix btrfs_next_leaf to check for new items after dropping locks

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoFix btrfs_del_ordered_inode to allow forcing the drop during unlinks
Chris Mason [Wed, 25 Jun 2008 20:01:30 +0000 (16:01 -0400)]
Fix btrfs_del_ordered_inode to allow forcing the drop during unlinks

This allows us to delete an unlinked inode with dirty pages from the list
instead of forcing commit to write these out before deleting the inode.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoDrop locks in btrfs_search_slot when reading a tree block.
Chris Mason [Wed, 25 Jun 2008 20:01:30 +0000 (16:01 -0400)]
Drop locks in btrfs_search_slot when reading a tree block.

One lock per btree block can make for significant congestion if everyone
has to wait for IO at the high levels of the btree.  This drops
locks held by a path when doing reads during a tree search.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Replace the big fs_mutex with a collection of other locks
Chris Mason [Wed, 25 Jun 2008 20:01:30 +0000 (16:01 -0400)]
Btrfs: Replace the big fs_mutex with a collection of other locks

Extent alloctions are still protected by a large alloc_mutex.
Objectid allocations are covered by a objectid mutex
Other btree operations are protected by a lock on individual btree nodes

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Start btree concurrency work.
Chris Mason [Wed, 25 Jun 2008 20:01:30 +0000 (16:01 -0400)]
Btrfs: Start btree concurrency work.

The allocation trees and the chunk trees are serialized via their own
dedicated mutexes.  This means allocation location is still not very
fine grained.

The main FS btree is protected by locks on each block in the btree.  Locks
are taken top / down, and as processing finishes on a given level of the
tree, the lock is released after locking the lower level.

The end result of a search is now a path where only the lowest level
is locked.  Releasing or freeing the path drops any locks held.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add a thread pool just for submit_bio
Chris Mason [Thu, 12 Jun 2008 18:46:17 +0000 (14:46 -0400)]
Btrfs: Add a thread pool just for submit_bio

If a bio submission is after a lock holder waiting for the bio
on the work queue, it is possible to deadlock.  Move the bios
into their own pool.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBTRFS_IOC_TRANS_START should be privilegued
Christoph Hellwig [Thu, 12 Jun 2008 01:53:58 +0000 (21:53 -0400)]
BTRFS_IOC_TRANS_START should be privilegued

As mentioned in the comment next to it btrfs_ioctl_trans_start can
do bad damage to filesystems and thus should be limited to privilegued
users.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: split out ioctl.c
Christoph Hellwig [Thu, 12 Jun 2008 01:53:53 +0000 (21:53 -0400)]
Btrfs: split out ioctl.c

Split the ioctl handling out of inode.c into a file of it's own.
Also fix up checkpatch.pl warnings for the moved code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: kerneldoc comments for extent_map.c
Christoph Hellwig [Thu, 12 Jun 2008 01:52:17 +0000 (21:52 -0400)]
Btrfs: kerneldoc comments for extent_map.c

Add kerneldoc comments for all exported functions.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add a mount option to control worker thread pool size
Chris Mason [Thu, 12 Jun 2008 01:47:56 +0000 (21:47 -0400)]
Btrfs: Add a mount option to control worker thread pool size

mount -o thread_pool_size changes the default, which is
min(num_cpus + 2, 8).  Larger thread pools would make more sense on
very large disk arrays.

This mount option controls the max size of each thread pool.  There
are multiple thread pools, so the total worker count will be larger
than the mount option.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Worker thread optimizations
Chris Mason [Thu, 12 Jun 2008 00:21:24 +0000 (20:21 -0400)]
Btrfs: Worker thread optimizations

This changes the worker thread pool to maintain a list of idle threads,
avoiding a complex search for a good thread to wake up.

Threads have two states:

idle - we try to reuse the last thread used in hopes of improving the batching
ratios

busy - each time a new work item is added to a busy task, the task is
rotated to the end of the line.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add backport for the kthread work on kernels older than 2.6.20
Chris Mason [Wed, 11 Jun 2008 21:09:48 +0000 (17:09 -0400)]
Btrfs: Add backport for the kthread work on kernels older than 2.6.20

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fix mount -o max_inline=0
Chris Mason [Wed, 11 Jun 2008 20:51:38 +0000 (16:51 -0400)]
Btrfs: Fix mount -o max_inline=0

max_inline=0 used to force the max_inline size to one sector instead.  Now
it properly disables inline data items, while still being able to read
any that happen to exist on disk.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add async worker threads for pre and post IO checksumming
Chris Mason [Wed, 11 Jun 2008 20:50:36 +0000 (16:50 -0400)]
Btrfs: Add async worker threads for pre and post IO checksumming

Btrfs has been using workqueues to spread the checksumming load across
other CPUs in the system.  But, workqueues only schedule work on the
same CPU that queued the work, giving them a limited benefit for systems with
higher CPU counts.

This code adds a generic facility to schedule work with pools of kthreads,
and changes the bio submission code to queue bios up.  The queueing is
important to make sure large numbers of procs on the system don't
turn streaming workloads into random workloads by sending IO down
concurrently.

The end result of all of this is much higher performance (and CPU usage) when
doing checksumming on large machines.  Two worker pools are created,
one for writes and one for endio processing.  The two could deadlock if
we tried to service both from a single pool.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agobtrfs: allow scanning multiple devices during mount
Christoph Hellwig [Tue, 10 Jun 2008 14:40:46 +0000 (10:40 -0400)]
btrfs: allow scanning multiple devices during mount

Allows to specify one or multiple device=/dev/foo options during mount
so that ioctls on the control device can be avoided.  Especially useful
when trying to mount a multi-device setup as root.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agobtrfs: sanity mount option parsing and early mount code
Christoph Hellwig [Tue, 10 Jun 2008 14:40:29 +0000 (10:40 -0400)]
btrfs: sanity mount option parsing and early mount code

Also adds lots of comments to describe what's going on here.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agobtrfs: fix strange indentation in lookup_extent_mapping
Christoph Hellwig [Tue, 10 Jun 2008 14:21:04 +0000 (10:21 -0400)]
btrfs: fix strange indentation in lookup_extent_mapping

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agobtrfs: tiny makefile cleanup
Christoph Hellwig [Tue, 10 Jun 2008 14:20:57 +0000 (10:20 -0400)]
btrfs: tiny makefile cleanup

use normal kbuild syntax to build acl.o conditinally and remove comment
out lines.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: transaction ioctls
Sage Weil [Tue, 10 Jun 2008 14:07:39 +0000 (10:07 -0400)]
Btrfs: transaction ioctls

These ioctls let a user application hold a transaction open while it
performs a series of operations.  A final ioctl does a sync on the fs
(closing the current transaction).  This is the main requirement for
Ceph's OSD to be able to keep the data it's storing in a btrfs volume
consistent, and AFAICS it works just fine.  The application would do
something like

fd = ::open("some/file", O_RDONLY);
::ioctl(fd, BTRFS_IOC_TRANS_START);
/* do a bunch of stuff */
::ioctl(fd, BTRFS_IOC_TRANS_END);
or just
::close(fd);

And to ensure it commits to disk,

::ioctl(fd, BTRFS_IOC_SYNC);

When a transaction is held open, the trans_handle is attached to the
struct file (via private_data) so that it will get cleaned up if the
process dies unexpectedly.  A held transaction is also ended on fsync() to
avoid a deadlock.

A misbehaving application could also deliberately hold a transaction open,
effectively locking up the FS, so it may make sense to restrict something
like this to root or something.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Dislable acl xattr handlers
Yan [Tue, 10 Jun 2008 02:21:46 +0000 (22:21 -0400)]
Btrfs: Dislable acl xattr handlers

The acl code is not yet complete, and the xattr handlers are causing
problems for cp -p on some distros.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: bdi_init and bdi_destroy come with 2.6.23
Jan Engelhardt [Tue, 10 Jun 2008 02:19:40 +0000 (22:19 -0400)]
Btrfs: bdi_init and bdi_destroy come with 2.6.23

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agobtrfsctl -A error code fixup
Linda Knippers [Tue, 10 Jun 2008 02:17:11 +0000 (22:17 -0400)]
btrfsctl -A error code fixup

Send the error back to userland if the ioctl fails

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Invalidate dcache entry after creating snapshot and
Sven Wegener [Tue, 10 Jun 2008 01:57:42 +0000 (21:57 -0400)]
Btrfs: Invalidate dcache entry after creating snapshot and

We need to invalidate an existing dcache entry after creating a new
snapshot or subvolume, because a negative dache entry will stop us from
accessing the new snapshot or subvolume.

---
  ctree.h       |   23 +++++++++++++++++++++++
  inode.c       |    4 ++++
  transaction.c |    4 ++++
  3 files changed, 31 insertions(+)

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fix race in running_transaction checks
Chris Mason [Mon, 9 Jun 2008 13:35:50 +0000 (09:35 -0400)]
Btrfs: Fix race in running_transaction checks

When a new transaction was started, the code would incorrectly
set the pointer in fs_info before all the data structures were setup.
fsync heavy workloads hit races on the setup of the ordered inode spinlock

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agobtrfs delete ordered inode handling fix
Mingming [Tue, 27 May 2008 14:55:43 +0000 (10:55 -0400)]
btrfs delete ordered inode handling fix

Use btrfs_release_file instead of a put_inode call

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Always use the async submission queue for checksummed writes
Chris Mason [Tue, 27 May 2008 14:52:17 +0000 (10:52 -0400)]
Btrfs: Always use the async submission queue for checksummed writes

This avoids IO stalls and poorly ordered IO from inline writers mixing in
with the async submission queue

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Allocator fix variety pack
Chris Mason [Sat, 24 May 2008 18:04:53 +0000 (14:04 -0400)]
Btrfs: Allocator fix variety pack

* Force chunk allocation when find_free_extent has to do a full scan
* Record the max key at the start of defrag so it doesn't run forever
* Block groups might not be contiguous, make a forward search for the
  next block group in extent-tree.c
* Get rid of extra checks for total fs size
* Fix relocate_one_reference to avoid relocating the same file data block
  twice when referenced by an older transaction
* Use the open device count when allocating chunks so that we don't
  try to allocate from devices that don't exist

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Use kzalloc on the fs_devices allocation
Chris Mason [Fri, 16 May 2008 17:30:15 +0000 (13:30 -0400)]
Btrfs: Use kzalloc on the fs_devices allocation

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Handle transid == 0 while opening devices
Chris Mason [Fri, 16 May 2008 17:14:57 +0000 (13:14 -0400)]
Btrfs: Handle transid == 0 while opening devices

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Enable btree balancing on old kernels again
Chris Mason [Fri, 16 May 2008 17:06:51 +0000 (13:06 -0400)]
Btrfs: Enable btree balancing on old kernels again

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Change the congestion functions to meter the number of async submits as well
Chris Mason [Thu, 15 May 2008 20:15:45 +0000 (16:15 -0400)]
Btrfs: Change the congestion functions to meter the number of async submits as well

The async submit workqueue was absorbing too many requests, leading to long
stalls where the async submitters were stalling.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoFix corners in writepage and btrfs_truncate_page
Chris Mason [Thu, 15 May 2008 13:13:45 +0000 (09:13 -0400)]
Fix corners in writepage and btrfs_truncate_page

The extent_io writepage calls needed an extra check for discarding
pages that started on th last byte in the file.

btrfs_truncate_page needed checks to make sure the page was still part
of the file after reading it, and most importantly, needed to wait for
all IO to the page to finish before freeing the corresponding extents on
disk.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoFix btrfs_open_devices to deal with changes since the scan ioctls
Chris Mason [Tue, 13 May 2008 20:03:06 +0000 (16:03 -0400)]
Fix btrfs_open_devices to deal with changes since the scan ioctls

Devices can change after the scan ioctls are done, and btrfs_open_devices
needs to be able to verify them as they are opened and used by the FS.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add mount -o degraded to allow mounts to continue with missing devices
Chris Mason [Tue, 13 May 2008 17:46:40 +0000 (13:46 -0400)]
Btrfs: Add mount -o degraded to allow mounts to continue with missing devices

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Handle write errors on raid1 and raid10
Chris Mason [Mon, 12 May 2008 17:39:03 +0000 (13:39 -0400)]
Btrfs: Handle write errors on raid1 and raid10

When duplicate copies exist, writes are allowed to fail to one of those
copies.  This changeset includes a few changes that allow the FS to
continue even when some IOs fail.

It also adds verification of the parent generation number for btree blocks.
This generation is stored in the pointer to a block, and it ensures
that missed writes to are detected.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Pass down the expected generation number when reading tree blocks
Chris Mason [Mon, 12 May 2008 16:59:19 +0000 (12:59 -0400)]
Btrfs: Pass down the expected generation number when reading tree blocks

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Don't do btree balance_dirty_pages on old kernels, it stalls forever
Chris Mason [Fri, 9 May 2008 15:52:25 +0000 (11:52 -0400)]
Btrfs: Don't do btree balance_dirty_pages on old kernels, it stalls forever

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Chunk relocation fine tuning, and add a few printks to show progress
Chris Mason [Fri, 9 May 2008 15:46:48 +0000 (11:46 -0400)]
Btrfs: Chunk relocation fine tuning, and add a few printks to show progress

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: A number of nodatacow fixes
Chris Mason [Thu, 8 May 2008 20:31:21 +0000 (16:31 -0400)]
Btrfs: A number of nodatacow fixes

Once part of a delalloc request fails the cow checks, just cow the
entire range

It is possible for the back references to all be from the same root,
but still have snapshots against an extent.  The checks are now more strict,
forcing cow any time there are multiple refs against the data extent.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Only open block devices once during mount -o subvol=
Chris Mason [Thu, 8 May 2008 19:05:58 +0000 (15:05 -0400)]
Btrfs: Only open block devices once during mount -o subvol=

btrfs_open_devices needed a check to see if the device was already
open.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Update nodatacow mode to support cloned single files and resizing
Chris Mason [Thu, 8 May 2008 18:11:56 +0000 (14:11 -0400)]
Btrfs: Update nodatacow mode to support cloned single files and resizing

Before, nodatacow only checked to make sure multiple roots didn't have
references on a single extent.  This check makes sure that multiple
inodes don't have references.

nodatacow needed an extra check to see if the block group was currently
readonly.  This way cows forced by the chunk relocation code are honored.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Properly find the root for snapshotted blocks during chunk relocation
Chris Mason [Thu, 8 May 2008 17:26:18 +0000 (13:26 -0400)]
Btrfs: Properly find the root for snapshotted blocks during chunk relocation

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add support for online device removal
Chris Mason [Wed, 7 May 2008 15:43:44 +0000 (11:43 -0400)]
Btrfs: Add support for online device removal

This required a few structural changes to the code that manages bdev pointers:

The VFS super block now gets an anon-bdev instead of a pointer to the
lowest bdev.  This allows us to avoid swapping the super block bdev pointer
around at run time.

The code to read in the super block no longer goes through the extent
buffer interface.  Things got ugly keeping the mapping constant.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fix clone ioctl to not hold the path over inserts
Chris Mason [Mon, 5 May 2008 10:26:21 +0000 (06:26 -0400)]
Btrfs: Fix clone ioctl to not hold the path over inserts

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Silence bogus inode.c compiler warnings
Chris Mason [Fri, 2 May 2008 20:13:49 +0000 (16:13 -0400)]
Btrfs: Silence bogus inode.c compiler warnings

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add workaround for AppArmor changing remove_suid()
Jeff Mahoney [Fri, 2 May 2008 19:03:58 +0000 (15:03 -0400)]
Btrfs: Add workaround for AppArmor changing remove_suid()

In openSUSE 10.3, AppArmor modifies remove_suid to take a struct path
rather than just a dentry. This patch tests that the kernel is openSUSE
10.3 or newer and adjusts the call accordingly.

Debian/Ubuntu with AppArmor applied will also need a similar patch.
Maintainers of btrfs under those distributions should build on this
patch or, alternatively, alter their package descriptions to add
-DREMOVE_SUID_PATH to the compiler command line.

Signed-off-by: Jeff Mahoney <jeffm@suse.com>
- --- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ b/compat.h 2008-02-06 16:46:13.000000000 -0500
@@ -0,0 +1,15 @@
+#ifndef _COMPAT_H_
+#define _COMPAT_H_
+
+
+/*
+ * Even if AppArmor isn't enabled, it still has different prototypes.
+ * Add more distro/version pairs here to declare which has AppArmor applied.
+ */
+#if defined(CONFIG_SUSE_KERNEL)
+# if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22)
+# define REMOVE_SUID_PATH 1
+# endif
+#endif
+
+#endif /* _COMPAT_H_ */
- --- a/file.c 2008-02-06 11:37:39.000000000 -0500
+++ b/file.c 2008-02-06 16:46:23.000000000 -0500
@@ -37,6 +37,7 @@
 #include "ordered-data.h"
 #include "ioctl.h"
 #include "print-tree.h"
+#include "compat.h"

 static int btrfs_copy_from_user(loff_t pos, int num_pages, int write_bytes,
@@ -790,7 +791,11 @@ static ssize_t btrfs_file_write(struct f
  goto out_nolock;
  if (count == 0)
  goto out_nolock;
+#ifdef REMOVE_SUID_PATH
+ err = remove_suid(&file->f_path);
+#else
  err = remove_suid(fdentry(file));
+#endif
  if (err)
  goto out_nolock;
  file_update_time(file);

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fix do_sync_file_range ifdefs (2.6.22)
Chris Mason [Fri, 2 May 2008 18:49:33 +0000 (14:49 -0400)]
Btrfs: Fix do_sync_file_range ifdefs (2.6.22)

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Compile warning fixup in volume.c
Chris Mason [Fri, 2 May 2008 18:43:15 +0000 (14:43 -0400)]
Btrfs: Compile warning fixup in volume.c

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Clone file data ioctl
Sage Weil [Fri, 2 May 2008 18:43:14 +0000 (14:43 -0400)]
Btrfs: Clone file data ioctl

Add a new ioctl to clone file data

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fixes for 2.6.18 enterprise kernels
Chris Mason [Wed, 30 Apr 2008 17:59:35 +0000 (13:59 -0400)]
Btrfs: Fixes for 2.6.18 enterprise kernels

2.6.18 seems to get caught in an infinite loop when
cancel_rearming_delayed_workqueue is called more than once, so this switches
to cancel_delayed_work, which is arguably more correct.

Also, balance_dirty_pages can run into problems with 2.6.18 based kernels
because it doesn't have the per-bdi dirty limits.  This avoids calling
balance_dirty_pages on the btree inode unless there is actually something
to balance, which is a good optimization in general.

Finally there's a compile fix for ordered-data.h

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Tune stripe selection for raid1 and raid10
Chris Mason [Tue, 29 Apr 2008 18:12:09 +0000 (14:12 -0400)]
Btrfs: Tune stripe selection for raid1 and raid10

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Deal with failed writes in mirrored configurations
Chris Mason [Tue, 29 Apr 2008 13:38:00 +0000 (09:38 -0400)]
Btrfs: Deal with failed writes in mirrored configurations

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Drop some verbose printks
Chris Mason [Mon, 28 Apr 2008 20:40:52 +0000 (16:40 -0400)]
Btrfs: Drop some verbose printks

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add balance ioctl to restripe the chunks
Chris Mason [Mon, 28 Apr 2008 19:29:52 +0000 (15:29 -0400)]
Btrfs: Add balance ioctl to restripe the chunks

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add new ioctl to add devices
Chris Mason [Mon, 28 Apr 2008 19:29:42 +0000 (15:29 -0400)]
Btrfs: Add new ioctl to add devices

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Do more optimal file RA during shrinking and defrag
Chris Mason [Mon, 28 Apr 2008 13:02:36 +0000 (09:02 -0400)]
Btrfs: Do more optimal file RA during shrinking and defrag

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Avoid recursive chunk allocations
Chris Mason [Sat, 26 Apr 2008 15:03:32 +0000 (11:03 -0400)]
Btrfs: Avoid recursive chunk allocations

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Make the resizer work based on shrinking and growing devices
Chris Mason [Fri, 25 Apr 2008 20:53:30 +0000 (16:53 -0400)]
Btrfs: Make the resizer work based on shrinking and growing devices

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: write_cache_pages came in 2.6.22
Chris Mason [Fri, 25 Apr 2008 13:10:45 +0000 (09:10 -0400)]
Btrfs: write_cache_pages came in 2.6.22

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add failure handling for read_sys_array
Chris Mason [Fri, 25 Apr 2008 13:04:37 +0000 (09:04 -0400)]
Btrfs: Add failure handling for read_sys_array

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: write_extent_pages came in 2.6.23
Chris Mason [Fri, 25 Apr 2008 13:00:55 +0000 (09:00 -0400)]
Btrfs: write_extent_pages came in 2.6.23

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Throttle file_write when data=ordered is flushing the inode
Chris Mason [Fri, 25 Apr 2008 12:51:48 +0000 (08:51 -0400)]
Btrfs: Throttle file_write when data=ordered is flushing the inode

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fix balance_level to free the middle block if there is room in the left one
Chris Mason [Thu, 24 Apr 2008 18:42:46 +0000 (14:42 -0400)]
Btrfs: Fix balance_level to free the middle block if there is room in the left one

balance level starts by trying to empty the middle block, and then
pushes from the right to the middle.  This might empty the right block
and leave a small number of pointers in the middle.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Don't empty the middle buffer in push_nodes_for_insert
Chris Mason [Thu, 24 Apr 2008 14:54:32 +0000 (10:54 -0400)]
Btrfs: Don't empty the middle buffer in push_nodes_for_insert

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fix split_node to require more empty slots in the node as well
Chris Mason [Thu, 24 Apr 2008 13:34:34 +0000 (09:34 -0400)]
Btrfs: Fix split_node to require more empty slots in the node as well

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Make sure nodes have enough room for a double split
Chris Mason [Thu, 24 Apr 2008 13:22:51 +0000 (09:22 -0400)]
Btrfs: Make sure nodes have enough room for a double split

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fix the unplug_io_fn to grab a consistent copy of page->mapping
Chris Mason [Tue, 22 Apr 2008 17:26:47 +0000 (13:26 -0400)]
Btrfs: Fix the unplug_io_fn to grab a consistent copy of page->mapping

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoFix btrfs_get_extent and get_block corner cases, and disable O_DIRECT reads
Chris Mason [Tue, 22 Apr 2008 17:26:46 +0000 (13:26 -0400)]
Fix btrfs_get_extent and get_block corner cases, and disable O_DIRECT reads

The generic O_DIRECT code assumes all the bios have the same bdev,
which isn't true for multi-device btrfs.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Set nodatasum on the inode when written by a nodatasum mount
Chris Mason [Tue, 22 Apr 2008 13:24:20 +0000 (09:24 -0400)]
Btrfs: Set nodatasum on the inode when written by a nodatasum mount

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoDeal with page == NULL in the btrfs_unplug_io_fn
Chris Mason [Tue, 22 Apr 2008 13:22:11 +0000 (09:22 -0400)]
Deal with page == NULL in the btrfs_unplug_io_fn

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add a special device list for chunk allocations
Chris Mason [Tue, 22 Apr 2008 13:22:07 +0000 (09:22 -0400)]
Btrfs: Add a special device list for chunk allocations

This allows other code that needs to walk every device in the FS to do so
without locking against allocations.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Simplify device selection for mirrored reads
Chris Mason [Mon, 21 Apr 2008 16:01:38 +0000 (12:01 -0400)]
Btrfs: Simplify device selection for mirrored reads

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Make an unplug function that doesn't unplug every spindle
Chris Mason [Mon, 21 Apr 2008 14:03:05 +0000 (10:03 -0400)]
Btrfs: Make an unplug function that doesn't unplug every spindle

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Remove debugging statements from the invalidatepage calls
Chris Mason [Mon, 21 Apr 2008 12:52:50 +0000 (08:52 -0400)]
Btrfs: Remove debugging statements from the invalidatepage calls

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add 1MB to the min_free in alloc_chunk
Chris Mason [Mon, 21 Apr 2008 12:28:10 +0000 (08:28 -0400)]
Btrfs: Add 1MB to the min_free in alloc_chunk

This properly reflects the first 1MB we skip at the start of the device

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Scale the bdi ra_pages by the number of devices in the FS
Chris Mason [Fri, 18 Apr 2008 20:13:31 +0000 (16:13 -0400)]
Btrfs: Scale the bdi ra_pages by the number of devices in the FS

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoForce page->private removal in btrfs_invalidatepage
Chris Mason [Fri, 18 Apr 2008 20:11:30 +0000 (16:11 -0400)]
Force page->private removal in btrfs_invalidatepage

btrfs_invalidatepage is not allowed to leave pages around on the lru.
Any such pages will trigger an oops later on because the VM will see
page->private and assume it is a buffer head.

This also forces extra flushes of the async work queues before
dropping all the pages on the btree inode during unmount.  Left over
items on the work queues are one possible cause of busy state ranges
during truncate_inode_pages.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Set the btree inode i_size to OFFSET_MAX
Chris Mason [Fri, 18 Apr 2008 18:17:20 +0000 (14:17 -0400)]
Btrfs: Set the btree inode i_size to OFFSET_MAX

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Fix chunk allocation when some devices don't have enough room for stripes
Chris Mason [Fri, 18 Apr 2008 15:55:51 +0000 (11:55 -0400)]
Btrfs: Fix chunk allocation when some devices don't have enough room for stripes

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Calculate appropriate chunk sizes for both small and large filesystems
Chris Mason [Fri, 18 Apr 2008 14:29:51 +0000 (10:29 -0400)]
Btrfs: Calculate appropriate chunk sizes for both small and large filesystems

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Don't drop extent_map cache during releasepage on the btree inode
Chris Mason [Fri, 18 Apr 2008 14:29:50 +0000 (10:29 -0400)]
Btrfs: Don't drop extent_map cache during releasepage on the btree inode

The btree inode should only have a single extent_map in the cache,
it doesn't make sense to ever drop it.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add support for labels in the super block
Chris Mason [Fri, 18 Apr 2008 14:29:49 +0000 (10:29 -0400)]
Btrfs: Add support for labels in the super block

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Check device uuids along with devids
Chris Mason [Fri, 18 Apr 2008 14:29:38 +0000 (10:29 -0400)]
Btrfs: Check device uuids along with devids

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Remove bogus max_sector warnings from the extent_io code
Chris Mason [Thu, 17 Apr 2008 18:08:30 +0000 (14:08 -0400)]
Btrfs: Remove bogus max_sector warnings from the extent_io code

It was testing the bio before doing logical->physical mapping, so the
test was always wrong.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Avoid 64 bit div for RAID10
Chris Mason [Thu, 17 Apr 2008 15:58:30 +0000 (11:58 -0400)]
Btrfs: Avoid 64 bit div for RAID10

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Use the extent map cache to find the logical disk block during data retries
Chris Mason [Thu, 17 Apr 2008 15:29:12 +0000 (11:29 -0400)]
Btrfs: Use the extent map cache to find the logical disk block during data retries

The data read retry code needs to find the logical disk block before it
can resubmit new bios.  But, finding this block isn't allowed to take
the fs_mutex because that will deadlock with a number of different callers.

This changes the retry code to use the extent map cache instead, but
that requires the extent map cache to have the extent we're looking for.
This is a problem because btrfs_drop_extent_cache just drops the entire
extent instead of the little tiny part it is invalidating.

The bulk of the code in this patch changes btrfs_drop_extent_cache to
invalidate only a portion of the extent cache, and changes btrfs_get_extent
to deal with the results.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Only do async bio submission for pdflush
Chris Mason [Wed, 16 Apr 2008 17:06:16 +0000 (13:06 -0400)]
Btrfs: Only do async bio submission for pdflush

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Don't wait on tree block writeback before freeing them anymore
Chris Mason [Wed, 16 Apr 2008 16:59:22 +0000 (12:59 -0400)]
Btrfs: Don't wait on tree block writeback before freeing them anymore

This isn't required anymore because we don't reallocate blocks that
have already been written in this transaction.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Write bio checksumming outside the FS mutex
Chris Mason [Wed, 16 Apr 2008 15:15:20 +0000 (11:15 -0400)]
Btrfs: Write bio checksumming outside the FS mutex

This significantly improves streaming write performance by allowing
concurrency in the data checksumming.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Create a work queue for bio writes
Chris Mason [Wed, 16 Apr 2008 15:14:51 +0000 (11:14 -0400)]
Btrfs: Create a work queue for bio writes

This allows checksumming to happen in parallel among many cpus, and
keeps us from bogging down pdflush with the checksumming code.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add RAID10 support
Chris Mason [Wed, 16 Apr 2008 14:49:51 +0000 (10:49 -0400)]
Btrfs: Add RAID10 support

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: Add chunk uuids and update multi-device back references
Chris Mason [Tue, 15 Apr 2008 19:41:47 +0000 (15:41 -0400)]
Btrfs: Add chunk uuids and update multi-device back references

Block headers now store the chunk tree uuid

Chunk items records the device uuid for each stripes

Device extent items record better back refs to the chunk tree

Block groups record better back refs to the chunk tree

The chunk tree format has also changed.  The objectid of BTRFS_CHUNK_ITEM_KEY
used to be the logical offset of the chunk.  Now it is a chunk tree id,
with the logical offset being stored in the offset field of the key.

This allows a single chunk tree to record multiple logical address spaces,
upping the number of bytes indexed by a chunk tree from 2^64 to
2^128.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoBtrfs: A few updates for 2.6.18 and versions older than 2.6.25
Chris Mason [Mon, 14 Apr 2008 13:48:18 +0000 (09:48 -0400)]
Btrfs: A few updates for 2.6.18 and versions older than 2.6.25

This includes fixing a missing spinlock init call that caused oops on mount
for most kernels other than 2.6.25.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
16 years agoAdd a min size parameter to btrfs_alloc_extent
Chris Mason [Mon, 14 Apr 2008 13:46:10 +0000 (09:46 -0400)]
Add a min size parameter to btrfs_alloc_extent

On huge machines, delayed allocation may try to allocate massive extents.
This change allows btrfs_alloc_extent to return something smaller than
the caller asked for, and the data allocation routines will loop over
the allocations until it fills the whole delayed alloc.

Signed-off-by: Chris Mason <chris.mason@oracle.com>