Andi Kleen wrote:
On Wed, Mar 24, 2010 at 11:08:07PM -0400, jim owens wrote:
Agree, so the write design needs to prevent bad checksums.
How? Do you have a plan for that?
Yes... have Josef do it. ;)
The options I considered are:
1 - buffer always for uncompressed, the same as compressed.
2
sorry, this time with cc to list
Andi Kleen wrote:
On Tue, Mar 23, 2010 at 05:40:00PM -0400, jim owens wrote:
Andi Kleen wrote:
With checksums enabled, uncompressed reads aligned on the 4k block
are classic direct IO to user memory except at EOF.
Hmm, but what happens if the user
Andi Kleen wrote:
One thing that stroke me while reading is that, except for the out of line no
data
checksum case, this isn't real classical zero copy direct IO because
you always have to copy through some buffer.
Uh no, unless I really messed up or don't understand what you mean.
Direct I/O passes unaligned start and length values that must
be adjusted to block boundaries for lookup and merging when
an add_extent_mapping fails with EEXIST.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/inode.c | 56 -
1
Users of btrfs_lookup_first_ordered_extent() want to detect only a
real overlap, so searching for the range instead of just an address
prevents taking references on entries we don't want. This also
simplifies the code.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/file.c
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/dio.c | 22 ++
1 files changed, 14 insertions(+), 8 deletions(-)
diff --git a/fs/btrfs/dio.c b/fs/btrfs/dio.c
index b76b227..b6934be 100644
--- a/fs/btrfs/dio.c
+++ b/fs/btrfs/dio.c
@@ -486,8 +486,13 @@ getlock
This depends on the change to ordered data search.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/dio.c | 150 +++-
1 files changed, 104 insertions(+), 46 deletions(-)
diff --git a/fs/btrfs/dio.c b/fs/btrfs/dio.c
index b6934be
The 65k stripe length should be ignored as the stripes are
physically contiguous on disk so transfers can span stripes.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/volumes.c |6 +++---
1 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs
Returning a value greater than the caller's is ugly and prone
to dangerous future coding mistakes.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/extent-tree.c |3 ---
fs/btrfs/inode.c |5 ++---
fs/btrfs/volumes.c |9 +
3 files changed, 7 insertions
WRITE is needed on btrfs_map_block() to fetch all raid stripes,
READ only returns one device and we want to discard all copies.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/extent-tree.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/fs/btrfs/extent
The code in btrfs_unplug_io_fn() to selectively unplug just the
stripes affected by a page had been previously disabled. The
unneeded code for unplug_page in __btrfs_map_block made changes
for btrfs directio more difficult so I'm removing all of it.
Signed-off-by: jim owens owens6...@gmail.com
Compression and decompression can use a single z_stream and
z_stream.workspace memory as the operations always begin from
scratch and do not overlap in one thread's execution.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/zlib.c | 146
creation of the btrfs_multi_bio struct now calls
btrfs_map_to_stripe().
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/volumes.c | 85 +++
fs/btrfs/volumes.h | 10 ++
2 files changed, 62 insertions(+), 33 deletions(-)
diff --git a/fs
Direct I/O needs the total number of block devices and the
struct block_device for each to allocate control structures
prior to building bios for an extent. When building a bio,
the device's base physical address of the chunk is needed.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs
Use access_extent_buffer_page() to point at btree location of
inline compressed data so it can be inflated without a memcopy.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/extent_io.c | 16
fs/btrfs/extent_io.h |8 ++--
2 files changed, 22 insertions(+), 2
Direct I/O needs to efficiently fetch an extent range of checksums
so an option is added to copy raw checksums into a buffer instead
of locally allocating and returning btrfs_ordered_sum structs.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/ctree.h |2 +-
fs/btrfs/file
Direct I/O retry needs to know the number of mirror copies.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/volumes.c | 22 ++
fs/btrfs/volumes.h |1 +
2 files changed, 15 insertions(+), 8 deletions(-)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/compression.h | 20 +++
fs/btrfs/zlib.c| 142 +---
2 files changed, 154 insertions(+), 8 deletions(-)
diff --git a/fs/btrfs/compression.h b/fs/btrfs/compression.h
index 421f5b4
Direct I/O passes unaligned start and length values that must
be adjusted to block boundaries for lookup and merging when
an add_extent_mapping fails with EEXIST.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/inode.c | 56 -
1
Users of btrfs_lookup_first_ordered_extent() want to detect only a
real overlap, so searching for the range instead of just an address
prevents taking references on entries we don't want. This also
simplifies the code.
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/file.c
Signed-off-by: jim owens owens6...@gmail.com
---
fs/btrfs/Makefile |2 +-
fs/btrfs/inode.c |7 ++-
2 files changed, 3 insertions(+), 6 deletions(-)
diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile
index a35eb36..5cbe798 100644
--- a/fs/btrfs/Makefile
+++ b/fs/btrfs/Makefile
Chris,
Something that probably should be fixed is how
merging extent maps with block_len == -1 produces
illegal lengths, as in 8191.
I saw it with holes in directIO and it is not the
cause of my current problems so I'll hope someone
else decides to fix.
jim
--
To unsubscribe from this list:
The following patches add the field for tracking the smallest
device block size in the filesystem and using it instead of
the hard coded 512 byte values in dio.c.
I also implemented a simpler test for user misalignment on
devices with larger block sizes.
It passes fsx, but I have not tested
In a multi-device filesystem, it is possible to have devices with
different block sizes, as in 512, 1024, 2048, 4096. DirectIO read
will check user request alignment is valid for at least one device.
Signed-off-by: jim owens jim6...@gmail.com
---
fs/btrfs/volumes.c | 24
Instead of hard coding the minimum I/O alignment, use the smallest
bdev_logical_blocksize in the filesystem. Also change the alignment
tests to determine the real user request minimum alignment and make
all eof tail and device checks on that user blocksize.
Signed-off-by: jim owens jim6
Instead of hard coding the minimum I/O alignment, use the smallest
bdev_logical_blocksize in the filesystem. Also change the alignment
tests to determine the real user request minimum alignment and make
all eof tail and device checks on that user blocksize.
Signed-off-by: jim owens jim6
The code in btrfs_unplug_io_fn() to selectively unplug just the
stripes affected by a page had been previously disabled. The
unneeded code for unplug_page in __btrfs_map_block made changes
for btrfs directio more difficult so I'm removing all of it.
Signed-off-by: jim owens jow...@hp.com
Signed
Use access_extent_buffer_page() to point at btree location of
inline compressed data so it can be inflated without a memcopy.
Signed-off-by: jim owens jow...@hp.com
Signed-off-by: jim owens jim6...@gmail.com
---
V2 fixes whitespace checkpatch warning
fs/btrfs/extent_io.c | 16
Direct I/O needs to efficiently fetch an extent range of checksums
so an option is added to copy raw checksums into a buffer instead
of locally allocating and returning btrfs_ordered_sum structs.
Signed-off-by: jim owens jow...@hp.com
Signed-off-by: jim owens jim6...@gmail.com
---
V2 changes
Direct I/O retry needs to know the number of mirror copies.
Signed-off-by: jim owens jow...@hp.com
Signed-off-by: jim owens jim6...@gmail.com
---
V2 fixes whitespace checkpatch warning
fs/btrfs/volumes.c | 22 ++
fs/btrfs/volumes.h |1 +
2 files changed, 15
.
Signed-off-by: jim owens jow...@hp.com
Signed-off-by: jim owens jim6...@gmail.com
---
V2 fixes whitespace checkpatch warning and removes JIM testing printk
fs/btrfs/compression.h | 20 +++
fs/btrfs/zlib.c| 142 +---
2 files changed, 154
creation of the btrfs_multi_bio struct now calls
btrfs_map_to_stripe().
Signed-off-by: jim owens jow...@hp.com
Signed-off-by: jim owens jim6...@gmail.com
---
V2 fixes long line checkpatch warning
fs/btrfs/volumes.c | 85 +++
fs/btrfs/volumes.h
Use access_extent_buffer_page() to point at btree location of
inline compressed data so it can be inflated without a memcopy.
Signed-off-by: jim owens jow...@hp.com
Signed-off-by: jim owens jim6...@gmail.com
---
V2 fixes long line checkpatch warnings
fs/btrfs/extent_io.c | 16
Signed-off-by: jim owens jim6...@gmail.com
---
fs/btrfs/dio.c | 247
1 files changed, 142 insertions(+), 105 deletions(-)
diff --git a/fs/btrfs/dio.c b/fs/btrfs/dio.c
index 3315cc9..84277a9 100644
--- a/fs/btrfs/dio.c
+++ b/fs/btrfs
Signed-off-by: jim owens jim6...@gmail.com
---
fs/btrfs/dio.c | 24
1 files changed, 12 insertions(+), 12 deletions(-)
diff --git a/fs/btrfs/dio.c b/fs/btrfs/dio.c
index 84277a9..97ab77f 100644
--- a/fs/btrfs/dio.c
+++ b/fs/btrfs/dio.c
@@ -1487,8 +1487,8 @@ fail
The following 6 patches [dio.c V2 #/6] are code cleanup
from your review and my testing on a 32 bit laptop.
These patches do not have any changes to remove the
hard coded 511/512 or i_mutex. That will come later.
These patches are only to make it easier to see what
I have done, I will roll them
Signed-off-by: jim owens jim6...@gmail.com
---
fs/btrfs/dio.c | 23 ---
1 files changed, 12 insertions(+), 11 deletions(-)
diff --git a/fs/btrfs/dio.c b/fs/btrfs/dio.c
index 97ab77f..9f53577 100644
--- a/fs/btrfs/dio.c
+++ b/fs/btrfs/dio.c
@@ -199,7 +199,7 @@ static void
Signed-off-by: jim owens jim6...@gmail.com
---
fs/btrfs/dio.c | 72
1 files changed, 36 insertions(+), 36 deletions(-)
diff --git a/fs/btrfs/dio.c b/fs/btrfs/dio.c
index 9f53577..ad9b58f 100644
--- a/fs/btrfs/dio.c
+++ b/fs/btrfs/dio.c
Signed-off-by: jim owens jim6...@gmail.com
---
fs/btrfs/dio.c | 21 ++---
1 files changed, 10 insertions(+), 11 deletions(-)
diff --git a/fs/btrfs/dio.c b/fs/btrfs/dio.c
index ad9b58f..05a7445 100644
--- a/fs/btrfs/dio.c
+++ b/fs/btrfs/dio.c
@@ -127,7 +127,6 @@ struct
Signed-off-by: jim owens jim6...@gmail.com
---
fs/btrfs/dio.c |7 ++-
1 files changed, 2 insertions(+), 5 deletions(-)
diff --git a/fs/btrfs/dio.c b/fs/btrfs/dio.c
index 05a7445..b1beafc 100644
--- a/fs/btrfs/dio.c
+++ b/fs/btrfs/dio.c
@@ -459,7 +459,7 @@ getlock:
end
Hubert Kario wrote:
On Tuesday 02 March 2010 03:29:05 Robert Collins wrote:
As I say, I realise this is queued to get addressed anyway, but it seems
like a realistic thing for people to do (use BackupPC on btrfs) - even
if something better still can be written to replace the BackupPC store
in
Adam Kłobukowski wrote:
Hello
I'm running Ubuntu Lucid (2.6.32-14-generic-pae #20-Ubuntu SMP Sat Feb
20 07:07:46 UTC 2010 i686 GNU/Linux)
Doing online fsck.btrfs on 2TB volume dumps core:
fsck with the fs mounted is not supported.
there is a pending patch for fsck.btrfs to prevent it from
Chris Mason wrote:
[ use i_mutex for reads? ]
But, we already need the code that btrfs_page_mkwrite uses. It should
be enough to wait for the ordered extents and have the extent range
locked.
You don't mean have the lock_extent active while I issue the
btrfs_wait_ordered_range as I found
Josef Bacik wrote:
On Wed, Feb 10, 2010 at 01:53:50PM -0500, jim owens wrote:
+
+static int btrfs_dio_hole_read(struct btrfs_diocb *diocb, u64 hole_len)
+{
+ int err = 0;
+ diocb-umc.todo = hole_len;
+ while (diocb-umc.todo) {
+ struct bio_vec uv
Christoph Hellwig wrote:
On Mon, Feb 15, 2010 at 11:42:55AM -0500, Chris Mason wrote:
We can record the smallest alignment while scanning the devices and just
use that. In general 4K is good enough, although some other database
company might be more worried about smaller numbers.
How is
rk wrote:
http://www.osnews.com/story/22872/Linux_Not_Fully_Prepared_for_4096-Byte_Sector_Hard_Drives
Yes, partition alignment to the 4k drive block is critical,
but that is not an direct IO issue, it is mkfs.btrfs and
just as important for cache IO.
And in any case, many people will use 512
Christoph Hellwig wrote:
On Mon, Feb 15, 2010 at 05:26:34PM -0500, jim owens wrote:
My understanding is the current 4k drives normally operate in
512 byte read/write access mode unless you set them to run
as 4k only.
In 512 byte mode, they buffer internally on writes. It is probably
just
Josef Bacik wrote:
On Wed, Feb 10, 2010 at 01:53:50PM -0500, jim owens wrote:
+ /* traditional 512-byte device sector alignment is the
+* minimum required. if they have a larger sector disk
+* (possibly multiple sizes in the filesystem) and need
+ * a larger alignment
Signed-off-by: jim owens jow...@hp.com
Signed-off-by: jim owens owens6...@gmail.com
---
V2 is a merge of my original file:
http://article.gmane.org/gmane.comp.file-systems.btrfs/4530
and the fixes produced from Josef Bacik's fsx testing:
http://article.gmane.org/gmane.comp.file-systems.btrfs
RK wrote:
I think so too -- I have six 1TB drives on RAID-10 btrfs and it shows
that I have 5.5TB free space .. how that can be ?
# df -h
FilesystemSize Used Avail Use% Mounted on
/dev/sde1 66G 3.8G 59G 7% /
/dev/sda 5.5T 28K 5.5T 1%
Liuwenyi wrote:
Maybe, you should send this patch to
Andrew Morton a...@linux-foundation.org
Andrea Arcangeli aarca...@redhat.com
Michael S. Tsirkin m...@redhat.com
linux-ker...@vger.kernel.org
linux...@kvack.org
I said in [00] that I sent this patch to Andrew.
What I did not say was it
jim owens wrote:
What I did not say was it was sent on December 9, 2009
to every addess on your list, except for Andrea.
OK, as josef told me, the patch is corrupt and will not apply.
I am sending a new one I just made against btrfs-unstable
to linux-btrfs and will create a new patch
This is the completed code for direct I/O read. It handles all extent
types, validates checksums, and does retries.
I have tested all the paths (on X86-64) but it needs testing with
automation in a larger setup. If this design is acceptable,
Eric Whitney will run those tests.
The code supports
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/dio.c | 1902
1 files changed, 1902 insertions(+), 0 deletions(-)
create mode 100644 fs/btrfs/dio.c
diff --git a/fs/btrfs/dio.c b/fs/btrfs/dio.c
new file mode 100644
index 000
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/Makefile |2 +-
fs/btrfs/inode.c |7 ++-
2 files changed, 3 insertions(+), 6 deletions(-)
diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile
index a35eb36..5cbe798 100644
--- a/fs/btrfs/Makefile
+++ b/fs/btrfs/Makefile
@@ -7,4
Mitch Harder wrote:
On a single disk btrfs setup, such as for a desktop computer, what are
the implecations of creating your btrfs partition with '-m single'?
At first, I assumed I would want a single disk desktop setup to be
configured as 'single'. But that may not be the case for
.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/compression.h | 20 +++
fs/btrfs/zlib.c| 143 +---
2 files changed, 155 insertions(+), 8 deletions(-)
diff --git a/fs/btrfs/compression.h b/fs/btrfs/compression.h
index 421f5b4..afd262d
Steve Freitas wrote:
Hi all,
I was under the mistaken impression that btrfs checksumming, in its
current default configuration, protected your data from bitrot. It
appears this is not the case:
On Wed, 2010-01-06 at 18:24 +0100, Johannes Hirte wrote:
Am Mittwoch 06 Januar 2010 16:59:55
The 65k stripe length should be ignored as the stripes are
physically contiguous on disk so transfers can span stripes.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/volumes.c |6 +++---
1 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs
Returning a value greater than the caller's is ugly and prone
to dangerous future coding mistakes.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/extent-tree.c |3 ---
fs/btrfs/inode.c |5 ++---
fs/btrfs/volumes.c |9 +
3 files changed, 7 insertions(+), 10
creation of the btrfs_multi_bio struct now calls
btrfs_map_to_stripe().
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/volumes.c | 84 +++
fs/btrfs/volumes.h | 10 ++
2 files changed, 61 insertions(+), 33 deletions(-)
diff --git a/fs/btrfs
Direct I/O needs the total number of block devices and the
struct block_device for each to allocate control structures
prior to building bios for an extent. When building a bio,
the device's base physical address of the chunk is needed.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs
Use access_extent_buffer_page() to point at btree location of
inline compressed data so it can be inflated without a memcopy.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/extent_io.c | 16
fs/btrfs/extent_io.h |3 +++
2 files changed, 19 insertions(+), 0 deletions
Direct I/O needs to efficiently fetch an extent range of checksums
so an option is added to copy raw checksums into a buffer instead
of locally allocating and returning btrfs_ordered_sum structs.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/ctree.h |2 +-
fs/btrfs/file-item.c
As if a Monday after a long holiday was not bad enough... this series
has some simple patches in existing btrfs code that I want to make for
btrfs directio code.
The RESEND patches 1-5 were sent to Chris in December because IMO
they have value without my directio code. You need them to apply
Direct I/O needs to fetch a single checksum based on the
extent information instead of page struct content.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/ctree.h |2 ++
fs/btrfs/file-item.c | 45 +
2 files changed, 47 insertions(+), 0
The code in btrfs_unplug_io_fn() to selectively unplug just the
stripes affected by a page had been previously disabled. The
unneeded code for unplug_page in __btrfs_map_block made changes
for btrfs directio more difficult so I'm removing all of it.
Signed-off-by: jim owens jow...@hp.com
Christian Kujau wrote:
On 26.12.09 08:00, jim owens wrote:
I was using sync to make sure that the data should be on the disks
Good, but not good enough for many tests... info sync
[...]
On Linux, sync is only guaranteed to schedule the dirty blocks
for
writing; it can
The 65k stripe length should be ignored as the stripes are
physically contiguous on disk so transfers can span stripes.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/volumes.c |6 +++---
1 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs
Returning a value greater than the caller's is ugly and prone
to dangerous future coding mistakes.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/extent-tree.c |3 ---
fs/btrfs/inode.c |5 ++---
fs/btrfs/volumes.c |9 +
3 files changed, 7 insertions(+), 10
WRITE is needed on btrfs_map_block() to fetch all raid stripes,
READ only returns one device and we want to discard all copies.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/extent-tree.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/fs/btrfs/extent-tree.c b/fs
Andrey Kuzmin wrote:
On Tue, Nov 17, 2009 at 12:48 AM, jim owens jow...@hp.com wrote:
But as we have said many times... if we have different
raid types active on different files, any attempt to make
Late question, but could you please explain it a bit further (or point
me to respective
Leszek Ciesielski wrote:
Hi,
the results of running 'df' against a btrfs volume are somewhat
unintuitive from a user point of view. On a single drive btrfs volume,
created with 'mkfs.btrfs -m raid1 -d raid1 /dev/sda6', I am getting
the following result:
/dev/sda6 1.4T 594G 804G
Goffredo Baroncelli wrote:
I don't know a software which need so many hard links. But it easy to find
some similar cases.
For example under my /usr/bin I have 478 _soft links_ to _different_
files.
Hard link is not used in place of soft link... soft link is
a different and preferred
David Woodhouse wrote:
http://git.infradead.org/users/dwmw2/btrfs-raid56.git
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/volumes.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 95babc1..913c29f 100644
--- a/fs
Still not ready for review or for anyone else to try it,
this V2 update fixes the bugs I have found in my testing.
New things that now seem to work:
* multiple disks (btrfs raid 0,1,10,5,6)
* AIO
* multiple buffer vectors
* files that are not a multiple of 512 bytes
jim
Massimo Maggi wrote:
Hi,
SSDs have low latency but a high price per GB,
Traditional hard disks have high latency, but high sequential read/write
speed and low price per GB.
Is possibile to use a SSD for metadata, which requires many seeks and is
relatively small, in a special RAID mode with a
This is my still-working-on-it code for btrfs directIO read.
I'm posting it so people can see the progress being made on
the project and can take an early shot at telling me this is
just a bad idea and I'm crazy if they want to, or point out
where I made some stupid mistake with btrfs core
Arrrgh...
If I did not already scare you enough to stay away from trying
this code, I already hit another bug. Proving the axiom that
you will find a bad bug as soon as you send your code to anyone.
* non-inline file size must be a multiple of device sector size.
jim
--
To unsubscribe from
Roy Sigurd Karlsbakk wrote:
Currently, btrfs doesn't support using direct I/O for read.
Is anyone working on this? It /is/ a rather basic feature, and testing
with iozone would be nice.
Yes, I am working on it. I will be away for the next 10 days
but hope to have something for read tests
Andrey Kuzmin wrote:
On Tue, Apr 28, 2009 at 2:02 PM, Chris Mason chris.ma...@oracle.com wrote:
On Tue, 2009-04-28 at 07:22 +0200, Thomas Glanzmann wrote:
Hello Chris,
There is a btrfs ioctl to clone individual files, and this could be used
to implement an online dedup. But, since it is
I'm working on O_DIRECT support... but with checksumming.
I've also been looking at AIO but directio is first.
jim
--
To unsubscribe from this list: send the line unsubscribe linux-btrfs in
the body of a message to majord...@vger.kernel.org
More majordomo info at
ftrace does not exist in earlier kernels and requires backport patching
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/async-thread.c |1 -
fs/btrfs/delayed-ref.c |1 -
2 files changed, 0 insertions(+), 2 deletions(-)
diff --git a/fs/btrfs/async-thread.c b/fs/btrfs/async
CSights wrote:
Hi everyone,
I'm curious what would happen in btrfs if the following commands were
issued:
# cp file1 file2
# chown newuser:newgroup file2
Where file1 was owned by olduser:oldgroup.
If I understand copy-on-write correctly the cp would merely create a new
pointer (or
Lee Trager wrote:
Sorry it took so long to reply its been a crazy week for me.
You're lucky it was only a week, I often have crazy months!
I can't believe I made such a stupid mistake.
Back when I started in computers, I thought 16 bits would
be more than I would ever need to count the
Josef Bacik wrote:
On Mon, Mar 09, 2009 at 03:21:06PM -, Oliver Mattos wrote:
So the idea of the function is to return an integer in the range
[0,100]?
Why are we using a range of 0 to 100 anyway? 100 seems like an arbitary
value for kernel space - why not just keep it as a value in the
Lee,
The 02/26/2009 version is missing some of the patches that
were in the 02/20/2009 version so it will not compile on 26.
I found the bug in the patch that breaks on 26:
+#if LINUX_VERSION_CODE = KERNEL_VERSION(2, 6, 27)
if (!trylock_page(page))
+#else
+
Lee Trager wrote:
The more and more I look at this problem the more I tend to think that
the issue is because of some change in the way the VFS or something
interacts with the file system. Does anyone know of any big changes? Why
is the inode being marked dirty? Is there some kind of read error.
lee,
A couple of thoughts about your .26 lockup:
- I assume you are on the same hardware with 27.
- Are you using a module or builtin btrfs (I build it in).
- It looks like you are using vmware... when I'm doing kernel
stuff I want to be on the iron, not trusting virtual machine code.
Pavel Machek wrote:
If you don't want it, don't compile it in. The Kconfig text is very
clear.
No, I'd not expect that option to panic systems. That's why I
suggested:
diff --git a/fs/xfs/Kconfig b/fs/xfs/Kconfig
index 29228f5..b7ac847 100644
--- a/fs/xfs/Kconfig
+++ b/fs/xfs/Kconfig
@@
Add call to LSM security initialization and save
resulting security xattr for new inodes.
Add xattr support to symlink inode ops.
Set inode-i_op for existing special files.
Signed-off-by: jim owens jow...@hp.com
---
fs/btrfs/inode.c | 23 +++
fs/btrfs/xattr.c | 32
Josef Bacik wrote:
On Wed, Jan 07, 2009 at 03:19:38PM -0500, jim owens wrote:
+int btrfs_xattr_security_init(struct inode *inode, struct inode *dir)
+{
+ int err;
+ size_t len;
+ void *value;
+ char *suffix;
+ char *name;
+
+ err
Diego Calleja wrote:
El Tue, 9 Dec 2008 13:09:51 -0500, Lee Trager [EMAIL PROTECTED] escribió:
It does seem that doing it with volumes would limit user control and add
lots of complexity to such a simple task.
IMHO, WRT compression it's the contrary. Compression on a per-file basis has
never
Ric Wheeler wrote:
Matthias Wächter wrote:
On 10/22/2008 3:50 PM, Chris Mason wrote:
Let me reword my answer ;). The next write will always succeed unless
the drive is out of remapping sectors. If the drive is out, it is only
good for reads and holding down paper on your desk.
I
Michel Salim wrote:
Though it would be nice to have a tool that would provide enough
information to make a warranty claim -- does btrfs keep enough
information for such a tool to be written?
Failed device I/O (rather than bad checksums and other
fs-specific error detections) should be logged
Morey Roof wrote:
Could someone write a quick recap of this for those of us who are
interested and happened to miss it?
Everyone on the call seems to be waiting for someone else
to respond. I'll do it just to make up for dedup :)
- Chris said he saw major performance problems from pdflush
Morey Roof wrote:
Thank you. Also, did they get around to talking about when they are
going to merge in some of the queued up patches?
We have never discussed code merges on any calls.
Probably because Chris does all that merge work.
I would not expect anything until he finishes the flush
Improved allocator threading
I wanted to work on the allocator with a larger scope
where threading is only a minor part of trying to address
these items from the Project_ideas that I think could change
disk format in some way (to fix it before v1.0):
- Different sector sizes
- Multiple chunk
Chris Mason wrote:
Josef's allocator fix is on the list because we currently fall over in
some workloads at 100% cpu time when the FS is 60% full. The space
indexing is complex and strange, it just needs to be redone.
I don't understand. Do you mean josef wants someone to fix
multithreading
99 matches
Mail list logo