Btrfs_read_block_groups() function is the most time consuming function
if the whole fs is filled with small extents.
For a btrfs filled with all 16K sized files, and when 2T space is used,
mount the fs needs 10 to 12 seconds.
While ftrace shows that, btrfs_read_block_groups() takes about 9
second
On Wed, May 04, 2016 at 03:43:26PM -0400, Noah Massey wrote:
> When printing the countdown in the safety delay, the number should
> correspond to the number of seconds remaining to wait at the time the
> delay is printed.
>
> In other words, there should be a one second sleep after printing '1'.
>
On Thu, May 5, 2016 at 5:23 AM, Zygo Blaxell
wrote:
> During a mount, we start the cleaner kthread first because the transaction
> kthread wants to wake up the cleaner kthread. We start the transaction
> kthread next because everything in btrfs wants transactions. We do reloc
> recovery in the t
On giovedì 5 maggio 2016 03:07:37 CEST, Chris Murphy wrote:
I suggest using defaults for starters. The only thing in that list
that needs be there is either subvolid or subvold, not both. Add in
the non-default options once you've proven the defaults are working,
and add them one at a time.
Yes
On 2016-05-04 19:18, Dmitry Katsubo wrote:
Dear btrfs community,
I am interested in spare volumes and hot auto-replacement feature [1]. I have a
couple of questions:
* Which kernel version this feature will be included?
Probably 4.7. I would not suggest using it in production for at least a
Most of it (like policy tuning/configuring/notification) is through
sysfs interface, However to implement this, we need the existing sysfs
volume patches to be integrated.
We need to think about the implementation of per-FSID spare which I
hope will solve the problem incompatible spare disk.
As
Hi,
the compression workspaces are allocated as needed an this could fail if
there's no free memory. Moreover, as we might be flushing data from the
restricted contexts we should try our best not to fail.
This patchset preallocates one workspace for each compression type at module
load time (and
The names are confusing, pick more fitting names and add comments.
Signed-off-by: David Sterba
---
fs/btrfs/compression.c | 35 +++
1 file changed, 19 insertions(+), 16 deletions(-)
diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index ff61a41ac90b..
Be verbose if there are no workspaces at all, ie. the module init time
preallocation failed.
Signed-off-by: David Sterba
---
fs/btrfs/compression.c | 14 ++
1 file changed, 14 insertions(+)
diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index c70625560265..658c39b70fba
With just one preallocated workspace we can guarantee forward progress
even if there's no memory available for new workspaces. The cost is more
waiting but we also get rid of several error paths.
On average, there will be several idle workspaces, so the waiting
penalty won't be so bad.
In the wor
Preallocate one workspace for each compression type so we can guarantee
forward progress in the worst case. A failure cannot be a hard error as
we might not use compression at all on the filesystem. If we can't
allocate the workspaces later when need them, it might actually
deadlock, but in such si
From: Filipe Manana
The btrfs_rename_exchange() started as a copy-paste from btrfs_rename(),
which had a race fixed by my previous patch titled "Btrfs: pin log earlier
when renaming", and so it suffers from the same problem.
We pin the logs of the affected roots after we insert the new inode
ref
From: Filipe Manana
If rename exchange operations fail at some point after we pinned any of
the logs, we end up aborting the current transaction but never unpin the
logs, which leaves concurrent tasks that are trying to sync the logs (as
part of an fsync request from user space) blocked forever a
From: Filipe Manana
If we failed to fully setup the whiteout inode during a rename operation
with the whiteout flag, we ended up leaking the inode, not decrementing
its link count nor removing all its items from the fs/subvol tree.
Signed-off-by: Filipe Manana
---
fs/btrfs/inode.c | 12 ++-
On Thu, May 05, 2016 at 12:36:52PM +0200, Niccolò Belli wrote:
> On giovedì 5 maggio 2016 03:07:37 CEST, Chris Murphy wrote:
> > I suggest using defaults for starters. The only thing in that list
> > that needs be there is either subvolid or subvold, not both. Add in
> > the non-default options onc
Hi,
Building on the discussion "Exposing Extent Information to Userspace"
at LSF, this patchset offers the userspace definition, implementation,
and manpages for a new FS_IOC_GETFSMAPX ioctl that enables userspace
to query the filesystem for a map of every extent in a given range of
physical block
Document the new XFS_IOC_GETFSMAPX that returns the physical layout
of a (disk-based) filesystem.
(Yes, the leading 'X' needs to fall off...)
Signed-off-by: Darrick J. Wong
---
man2/ioctl_getfsmapx.2 | 253
1 file changed, 253 insertions(+)
cre
Introduce a new ioctl that uses the reverse mapping btree to return
information about the physical layout of the filesystem. This is
the xfsprogs side of things for userspace support.
Signed-off-by: Darrick J. Wong
---
libxfs/xfs_fs.h | 65 +++
Add a new command, 'fsmap', to xfs_io so that we can query the filesystem
extent map on a live filesystem.
Signed-off-by: Darrick J. Wong
---
io/Makefile |2
io/fsmap.c| 485 +
io/init.c |1
io/io.h |
On Wed, May 04 2016, Dave Chinner wrote:
> FWIW, I don't think making evict() non-blocking is going to be worth
> the effort here. Making memory reclaim wait on a priority ordered
> queue while asynchronous reclaim threads run reclaim as efficiently
> as possible and wakes waiters as it frees the
20 matches
Mail list logo
