Chris Mason [Tue, 29 Jan 2008 14:59:12 +0000 (09:59 -0500)]
Btrfs: extent_io and extent_state optimizations
The end_bio routines are changed to take a pointer to the extent state
struct, and the state tree is walked in order to set/clear appropriate
bits as IO completes. This greatly reduces the number of rbtree searches
done by the end_bio handlers, and reduces lock contention.
The extent_io releasepage function is changed to avoid expensive searches
for locked state.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Chris Mason [Thu, 24 Jan 2008 21:13:08 +0000 (16:13 -0500)]
Btrfs: Split the extent_map code into two parts
There is now extent_map for mapping offsets in the file to disk and
extent_io for state tracking, IO submission and extent_bufers.
The new extent_map code shifts from [start,end] pairs to [start,len], and
pushes the locking out into the caller. This allows a few performance
optimizations and is easier to use.
A number of extent_map usage bugs were fixed, mostly with failing
to remove extent_map entries when changing the file.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Yan [Tue, 22 Jan 2008 17:46:56 +0000 (12:46 -0500)]
Fix for btrfs_find_free_objectid
btrfs_find_free_objectid may return a used objectid due to arithmetic
underflow. This bug may happen when parameter 'root' is tree root, so
it may cause serious problems when creating snapshot or sub-volume.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Chris Mason [Tue, 8 Jan 2008 20:46:30 +0000 (15:46 -0500)]
Btrfs: Add data=ordered support
This forces file data extents down the disk along with the metadata that
references them. The current implementation is fairly simple, and just
writes out all of the dirty pages in an inode before the commit.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Chris Mason [Fri, 4 Jan 2008 14:34:54 +0000 (09:34 -0500)]
Properly call btrfs_search_slot while shrinking
The shrinking code used btrfs_next_leaf to find the next item, but
this does not cow the blocks it touches. This fix calls search_slot after
finding the next item to do appropriate cow and balancing.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Yan [Thu, 3 Jan 2008 19:14:39 +0000 (14:14 -0500)]
Btrfs: Properly handle overlapping extent in shrink_extent_tree
The patch fixes the overlapping extent issue in shrink_extent_tree.
It checks whether there is an overlapping extent by using
find_previous_extent. If there is an overlapping extent, it setups
key.objectid and cur_byte properly.
---
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Chris Mason [Tue, 18 Dec 2007 01:14:01 +0000 (20:14 -0500)]
Btrfs: Add mount option to turn off data cow
A number of workloads do not require copy on write data or checksumming.
mount -o nodatasum to disable checksums and -o nodatacow to disable
both copy on write and checksumming.
In nodatacow mode, copy on write is still performed when a given extent
is under snapshot.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Yan [Fri, 14 Dec 2007 16:14:42 +0000 (11:14 -0500)]
Off by one fix for btrfs_drop_extents
One of my old patches introduces a new bug to
btrfs_drop_extents(changeset 275). Inline extents are not truncated
properly when "extent_end == end", it can trigger the BUG_ON at
file.c:600. I hope I don't introduce new bug this time.
---
Signed-off-by: Chris Mason <chris.mason@oracle.com>
compiling btrfs into the kernel results in section mismatch warnings. __exit
functions are called where they are not allowed to. The attached patch fixes
this for me. Not sure if it is correct though.
Signed-off-by: Christian Hesse <mail@earthworm.de>
--
Regards,
Chris
Yan [Mon, 26 Nov 2007 15:58:13 +0000 (10:58 -0500)]
Btrfs: Properly update right_nritems in push_leaf_left
The codes that fixup the right leaf and the codes that dirty the
extnet buffer use the variable 'right_nritems' , both of them expect
'right_nritems' is the number of items in right leaf after the push.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Josef Bacik [Mon, 19 Nov 2007 15:18:17 +0000 (10:18 -0500)]
Btrfs: Make ACLs return EOPNOTSUPP for now
There was a slight problem with ACL's returning EINVAL when you tried to set an
ACL. This isn't correct, we should be returning EOPNOTSUPP, so I did a very
ugly thing and just commented everybody out and made them return EOPNOTSUPP.
This is only temporary, I'm going back to implement ACL's, but Chris wants to
push out a release so this will suffice for now.
Also Yan suggested setting reada to -1 in the delete case to enable backwards
readahead, and in the listxattr case I moved path->reada = 2; to after the if
(!path) check so we can avoid a possible null dereference. Thank you,
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Yan [Fri, 16 Nov 2007 19:57:09 +0000 (14:57 -0500)]
Btrfs: Add full_scan parameter to find_search_start
This patch adds a new parameter 'full_scan' to 'find_search_start',
thereby 'find_search_start' can know whether 'find_free_extent' is in
full scan phrase. I feel that 'find_search_start' should skip calling
'btrfs_find_block_group' when 'find_free_extent' is in full scan
phrase. In my test on a 2GB volume, Oops occurs when space usage is
about 76%. After apply the patch, Oops occurs when space usage is
near 100%.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Yan [Fri, 16 Nov 2007 19:57:08 +0000 (14:57 -0500)]
Btrfs: Add block group pinned accounting back
This patch adds a helper function 'update_pinned_extents' to
extent-tree.c. The usage of the helper function is similar to
'update_block_group', the last parameter of the function indicates
pin vs unpin.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
projects / linux-beck.git / log