similar to: [PATCH 1/2] Btrfs: remove bad fsync transaction check

Add missing mnt_drop_write to match the mnt_want_write in btrfs_ioctl_trans_start. Signed-off-by: Sage Weil <sage@newdream.net> --- fs/btrfs/ioctl.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 6228b69..69c4a07 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1059,6 +1059,8 @@ long btrfs_ioctl_trans_end(struct

Hi all, This is an alternative approach to atomic user transactions for btrfs. The old start/end ioctls suffer from some basic limitations, namely - We can''t properly reserve space ahead of time to avoid ENOSPC part way through the transaction, and - The process may die (seg fault, SIGKILL) part way through the transaction. Currently when that happens the partial transaction will

Greetings all, I see the following issue during snap creation under IO: Transaction commit calls btrfs_start_delalloc_inodes() that locks the delalloc_inodes list, fetches the first inode, unlocks the list, triggers btrfs_alloc_delalloc_work/btrfs_queue_worker for this inode and then locks the list again. Then it checks the head of the list again. In my case, this is always exactly the same

This patch finally gives us full proper -ENOSPC handling for btrfs. Now whenever you do a btrfs_start_transaction, you must specify the number of items you are planning to add/delete/modify. The worst case number of blocks that could be modified by changing that number of items will be calculated and checked against the amount of free space in the space_info where the root you are modifying

Hi everyone, This patchset is the latest approach I''m using for the Ceph storage daemon to keep track of which data has safely committed to disk. The basic idea is to not use the (problematic) user transaction ioctls at all. Instead, the daemon quiesces its own write requests, initiates an async snapshot, and then continues. The snapshot approach is nice because it provides rollback.

btrfs_ilookup is unused, which is good because a normal filesystem should never have to use ilookup anyway. Remove it. Signed-off-by: Christoph Hellwig <hch@lst.de> Index: btrfs-unstable/ctree.h =================================================================== --- btrfs-unstable.orig/ctree.h 2008-08-30 11:08:58.000000000 -0300 +++ btrfs-unstable/ctree.h 2008-08-30 11:09:10.000000000

Hello, Ok here are all my fsync changes, including the fix for the put_transaction stuff that I posted earlier. I figure it'll be easier to just commit it as one thing then each individual peice. The additional thing that I did was add a radix tree that tracks the current outstanding transactions. I also added a field to the in memory btrfs inode that keeps track of the last transaction

Hi all, following the thread about the BTRFS_IOC_TREE_SEARCH ioctl [1], I made a patch which try to address the problem of restarting the ioctl. In the current solution is the application during the restart of the search to fill the min_* fields in the "struct btrfs_ioctl_search_key". In general the values set are the last one returned "+1". But doing so we reduce the

Hi Chris, This is the extent of my current queue of Btrfs snapshot/subvol/commit stuff. Most of these were posted several months ago. Can be sent upstream during this merge window? Not having this functionality is becoming a bit of a roadblock for our efforts to keep the Ceph data in a consistent state. These patches are also available from

Hi Chris- Here''s a simpler version of the patch that drops the unrelated sync_fs stuff. thanks- sage The ''flushoncommit'' mount option forces any data dirtied by a write in a prior transaction to commit as part of the current commit. This makes the committed state a fully consistent view of the file system from the application''s perspective (i.e., it

After running ceph on XFS for some time, I decided to try btrfs again. Performance with the current "for-linux-min" branch and big metadata is much better. The only problem (?) I''m still seeing is a warning that seems to occur from time to time: [87703.784552] ------------[ cut here ]------------ [87703.789759] WARNING: at fs/btrfs/inode.c:2103

Hi, Recently, I made a btrfs to use. And I met slowness problem. Trying to diag it. I found this: 1. dd if=/dev/zero of=test count=1024 bs=1MB This is fast, at about 25MB/s, and reasonable iowait. 2. dd if=/dev/zero of=test count=1 bs=1GB This is pretty slow, at about 1.5MB/s, and 90%+ iowait, constantly. May I know why it works like this? Thanks. -- To unsubscribe from this list: send the

We calculate timeout (either 1 or MAX_SCHEDULE_TIMEOUT) based on whether num_writers > 1 or should_grow at the top of the loop. Then, much much later, we wait for that timeout if either num_writers or should_grow is true. However, it''s possible for a racing process (calling btrfs_end_transaction()) to decrement num_writers such that we wait forever instead of for 1. Fix this by

When using delalloc workers in a non-waiting way (like for enospc handling) we can end up not actually waiting for the dirty pages to be started if we have compression. We need to add an extra filemap flush to make sure any async extents that have started are actually moved along before returning. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> --- fs/btrfs/inode.c | 18

The freeze rwsem is taken by sb_start_intwrite() and dropped during the commit_ or end_transaction(). In the async case, that happens in a worker thread. Tell lockdep the calling thread is releasing ownership of the rwsem and the async thread is picking it up. Josef and I worked out a more complicated solution that made the async commit thread join and potentially get a later transaction, but

Does Fsync of a file on a ext3 fs mounted with "ordered" option(the default) result in flush the dirty data buffers in the fs that correspond to previous transactions. In other words, if I keep writing to file1 (lots of data), log something to file2, keep fsyncing file2 after every write - does this mean file1 data would be committed by fsyncs on file2. Please copy me on your replies

Hi, I have a question on fsync() and ext3's journaling modes. Assume that I call fsync(fd) on a file. If that file is in 'data=journal' mode, would the fsync() return once the data gets safely into the journal ? On the other hand, if that file is in 'data=writeback' mode, would the fsync() return only when the data gets safely into its actual location ? Any help is

If I want to make sure a file has really been deleted do I have to: 1. fsync the file 2. fsync the directory the file is in 3. both? Much thanks. -- Ben Escoto

I'm currently looking at the behaviour of various Linux filesystems as backends for the Cyrus IMAP server. Like many mail applications, Cyrus makes heavy use of the fsync() system call to ensure that data has hit stable storage before it acknowledges receipt of messages. At the moment fsync() appears to introduce lots of synchronous I/O as data is flushed all the way through to the filesystem

On Sun, 11 Apr 2010 18:03:00 +0300, Avi Kivity <avi@redhat.com> wrote: > On 04/09/2010 05:56 PM, Ben Gamari wrote: > > On Mon, 29 Mar 2010 00:08:58 +0200, Andi Kleen<andi@firstfloor.org> wrote: > > > >> Ben Gamari<bgamari.foss@gmail.com> writes: > >> ext4/XFS/JFS/btrfs should be better in this regard > >> > >> >