linux-stable.git/fs/xfs/xfs_rtalloc.c, branch linux-3.4.y

xfs: fix deadlock in xfs_rtfree_extent

2012-03-22T20:31:06+00:00

To fix the deadlock caused by repeatedly calling xfs_rtfree_extent

 - removed xfs_ilock() and xfs_trans_ijoin() from xfs_rtfree_extent(),
   instead added asserts that the inode is locked and has an inode_item
   attached to it.
 - in xfs_bunmapi() when dealing with an inode with the rt flag
   call xfs_ilock() and xfs_trans_ijoin() so that the
   reference count is bumped on the inode and attached it to the
   transaction before calling into xfs_bmap_del_extent, similar to
   what we do in xfs_bmap_rtalloc.

Signed-off-by: Kamal Dasu 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Ben Myers

xfs: simplify xfs_trans_ijoin* again

2011-10-12T02:15:08+00:00

There is no reason to keep a reference to the inode even if we unlock
it during transaction commit because we never drop a reference between
the ijoin and commit.  Also use this fact to merge xfs_trans_ijoin_ref
back into xfs_trans_ijoin - the third argument decides if an unlock
is needed now.

I'm actually starting to wonder if allowing inodes to be unlocked
at transaction commit really is worth the effort.  The only real
benefit is that they can be unlocked earlier when commiting a
synchronous transactions, but that could be solved by doing the
log force manually after the unlock, too.

Signed-off-by: Christoph Hellwig 
Signed-off-by: Alex Elder

xfs: rename xfs_bmapi to xfs_bmapi_write

2011-10-12T02:15:04+00:00

Now that all the read-only users of xfs_bmapi have been converted to
use xfs_bmapi_read(), we can remove all the read-only handling cases
from xfs_bmapi().

Once this is done, rename xfs_bmapi to xfs_bmapi_write to reflect
the fact it is for allocation only. This enables us to kill the
XFS_BMAPI_WRITE flag as well.

Also clean up xfs_bmapi_write to the style used in the newly added
xfs_bmapi_read/delay functions.

Signed-off-by: Dave Chinner 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Alex Elder

xfs: remove xfs_bmapi_single()

2011-10-12T02:15:03+00:00

Now we have xfs_bmapi_read, there is no need for xfs_bmapi_single().
Change the remaining caller over and kill the function.

Signed-off-by: Dave Chinner 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Alex Elder

xfs: Remove the macro XFS_BUF_PTR

2011-07-25T20:03:13+00:00

Remove the definition and usages of the macro XFS_BUF_PTR.

Signed-off-by: Chandra Seetharaman 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Alex Elder

xfs: Remove the macro XFS_BUF_ERROR and family

2011-07-25T19:57:46+00:00

Remove the definitions and usage of the macros XFS_BUF_ERROR,
XFS_BUF_GETERROR and XFS_BUF_ISERROR.

Signed-off-by: Chandra Seetharaman 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Alex Elder

xfs: Convert remaining cmn_err() callers to new API

2011-03-06T23:08:35+00:00

Once converted, kill the remainder of the cmn_err() interface.

Signed-off-by: Dave Chinner 
Reviewed-by: Alex Elder 
Reviewed-by: Christoph Hellwig

xfs: stop using xfs_trans_iget in the RT allocator

2011-02-23T02:30:21+00:00

During mount we establish references to the RT inodes, which we keep for
the lifetime of the filesystem.  Instead of using xfs_trans_iget to grab
additional references when adding RT inodes to transactions use the
combination of xfs_ilock and xfs_trans_ijoin_ref, which archives the same
end result with less overhead.

Signed-off-by: Christoph Hellwig 
Reviewed-by: Dave Chinner 
Signed-off-by: Alex Elder

xfs: add lockdep annotations for the rt inodes

2011-02-07T19:29:18+00:00

The rt bitmap and summary inodes do not participate in the normal inode
locking protocol.  Instead the rt bitmap inode can be locked in any
transaction involving rt allocations, and the both of the rt inodes can
be locked at the same time.  Add specific lockdep subclasses for the rt
inodes to prevent lockdep from blowing up.

Signed-off-by: Christoph Hellwig 
Signed-off-by: Alex Elder

xfs: only lock the rt bitmap inode once per allocation

2011-02-07T19:29:06+00:00

Currently both xfs_rtpick_extent and xfs_rtallocate_extent call
xfs_trans_iget to grab and lock the rt bitmap inode, which results in a
deadlock since the removal of the lock recursion counters in commit

	"xfs: simplify inode to transaction joining"

Fix this by acquiring and locking the inode in xfs_bmap_rtalloc before
calling into xfs_rtpick_extent and xfs_rtallocate_extent.

Signed-off-by: Christoph Hellwig 
Signed-off-by: Alex Elder