From: Omar Sandoval
These should not be allowed.
Signed-off-by: Omar Sandoval
---
tests/generic/493 | 73 +++
tests/generic/493.out | 8 +
tests/generic/group | 1 +
3 files changed, 82 insertions(+)
create mode 100755 tests/generic/493
From: Omar Sandoval
Commit 8c96cfbfe530 ("generic/35[67]: disable swapfile tests on Btrfs")
disabled the swapfile tests on Btrfs because it did not support
swapfiles at the time. Now that we're adding support, we want these
tests to run, but they don't. _require_scratch_swapf
On Tue, May 22, 2018 at 02:41:45PM +0800, Eryu Guan wrote:
> On Wed, May 16, 2018 at 01:38:46PM -0700, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > Commit 8c96cfbfe530 ("generic/35[67]: disable swapfile tests on Btrfs")
> > disabled the swapfile test
From: Omar Sandoval
If we have invalid flags set, when we error out we must drop our writer
counter and free the buffer we allocated for the arguments. This bug is
trivially reproduced with the following program:
#include
#include
#include
#include
From: Omar Sandoval
In btrfs_extent_same(), we must check the NODATASUM flag while the
inodes are locked. Otherwise, it's possible that btrfs_ioctl_setflags()
will change the flags after we check. This was correct until a recent
change.
Fixes: 0a38effca37c ("Btrfs: split btrfs_e
From: Omar Sandoval
In btrfs_clone_files(), we must check the NODATASUM flag while the
inodes are locked. Otherwise, it's possible that btrfs_ioctl_setflags()
will change the flags after we check and we can end up with a party
checksummed file.
Fixes: 0e7b824c4ef9 ("Btrfs: don
From: Omar Sandoval
Clone and dedupe both have checks to make sure that we're not mixing
checksummed and not checksummed extents in a single file. However, both
checks are racy; we need to have the inodes locked or else the flags can
change after we check. The clone check has always been
o reason, and it could deadlock with delayed iput's.
Reviewed-by: Omar Sandoval
> Signed-off-by: Josef Bacik
> ---
> fs/btrfs/inode.c | 26 --
> fs/btrfs/super.c | 1 -
> 2 files changed, 27 deletions(-)
>
> diff --git a/fs/btrfs/inode.c b
On Tue, May 22, 2018 at 10:41:11AM -0700, Omar Sandoval wrote:
> On Tue, May 22, 2018 at 10:37:14AM -0700, Omar Sandoval wrote:
> > On Tue, May 22, 2018 at 07:17:48PM +0200, David Sterba wrote:
> > > On Tue, May 22, 2018 at 09:47:58AM -0700, Omar Sandoval wrote:
> >
On Tue, May 22, 2018 at 10:37:14AM -0700, Omar Sandoval wrote:
> On Tue, May 22, 2018 at 07:17:48PM +0200, David Sterba wrote:
> > On Tue, May 22, 2018 at 09:47:58AM -0700, Omar Sandoval wrote:
> > > From: Omar Sandoval
> > >
> > > Jun Wu at Facebook reported
On Tue, May 22, 2018 at 07:17:48PM +0200, David Sterba wrote:
> On Tue, May 22, 2018 at 09:47:58AM -0700, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > Jun Wu at Facebook reported that an internal service was seeing a return
> > value of 1 from ftruncate() on B
On Wed, May 16, 2018 at 11:52:37AM +0800, robbieko wrote:
> From: Robbie Ko
>
> This idea is from direct io. By this patch, we can make the buffered
> write parallel, and improve the performance and latency. But because we
> can not update isize without i_mutex, the unlocked buffered write just
>
From: Omar Sandoval
btrfs_truncate() uses two variables for error handling, ret and err (if
this sounds familiar, it's because btrfs_truncate_inode_items() did
something similar). This is error prone, as was made evident by "Btrfs:
fix error handling in btrfs_truncate()". W
From: Omar Sandoval
Jun Wu at Facebook reported that an internal service was seeing a return
value of 1 from ftruncate() on Btrfs in some cases. This is coming from
the NEED_TRUNCATE_BLOCK return value from btrfs_truncate_inode_items().
btrfs_truncate() uses two variables for error handling
On Mon, May 21, 2018 at 05:07:19PM -0700, Omar Sandoval wrote:
> From: Omar Sandoval
>
> I got a report that after upgrading to 4.16, someone's filesystems
> weren't mounting:
>
> [ 23.845852] BTRFS info (device loop0): unrecognized mount option 'subvol=&#x
From: Omar Sandoval
I got a report that after upgrading to 4.16, someone's filesystems
weren't mounting:
[ 23.845852] BTRFS info (device loop0): unrecognized mount option 'subvol='
Before 4.16, this mounted the default subvolume. It turns out that this
empty &quo
On Fri, May 18, 2018 at 04:26:16PM -0600, Chris Murphy wrote:
> On Fri, May 18, 2018 at 12:33 PM, Austin S. Hemmelgarn
> wrote:
> > On 2018-05-18 13:18, Niccolò Belli wrote:
> >>
> >> On venerdì 18 maggio 2018 19:10:02 CEST, Austin S. Hemmelgarn wrote:
> >>>
> >>> and also forces the people who ha
On Fri, May 18, 2018 at 02:43:02PM -0700, Omar Sandoval wrote:
> From: Omar Sandoval
>
> Jun Wu at Facebook reported that an internal service was seeing a return
> value of 1 from ftruncate() on Btrfs when compression is enabled. This
> is coming from the NEED_TRUNCATE_BLOCK re
From: Omar Sandoval
Jun Wu at Facebook reported that an internal service was seeing a return
value of 1 from ftruncate() on Btrfs when compression is enabled. This
is coming from the NEED_TRUNCATE_BLOCK return value from
btrfs_truncate_inode_items().
btrfs_truncate() uses two variables for
On Fri, May 18, 2018 at 05:14:35PM +0200, David Sterba wrote:
> On Fri, May 11, 2018 at 01:13:28PM -0700, Omar Sandoval wrote:
> > This is the fourth (and hopefully final) version of the orphan item
> > early ENOSPC and related fixes.
> >
> > Changes since v3:
>
From: Omar Sandoval
Commit 8c96cfbfe530 ("generic/35[67]: disable swapfile tests on Btrfs")
disabled the swapfile tests on Btrfs because it did not support
swapfiles at the time. Now that we're adding support, we want these
tests to run, but they don't. _require_scratch_swapf
From: Omar Sandoval
I'm going to add a bunch of tests for swap files, so create a group for
them and add the existing tests.
Signed-off-by: Omar Sandoval
---
tests/generic/group | 4 ++--
tests/xfs/group | 2 +-
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/tests/ge
From: Omar Sandoval
Changes since v1:
- Add patch 1 to create a group for swap
- Add a helper for formatting a swap file instead of open coding
everywhere
- Use $CHATTR_PROG instead of chattr in a few places that I forgot
Thanks!
Omar Sandoval (5):
xfstests: create swap group
generic
From: Omar Sandoval
These should not be allowed.
Signed-off-by: Omar Sandoval
---
tests/generic/489 | 73 +++
tests/generic/489.out | 8 +
tests/generic/group | 1 +
3 files changed, 82 insertions(+)
create mode 100755 tests/generic/489
From: Omar Sandoval
Swap files cannot have holes, and they must at least two pages.
swapon(8) and mkswap(8) have stricter restrictions, so add versions of
those commands without any restrictions.
Signed-off-by: Omar Sandoval
---
.gitignore| 2 ++
src/Makefile | 2
From: Omar Sandoval
Similar to generic/356 that makes sure we can't reflink an active
swapfile.
Signed-off-by: Omar Sandoval
---
tests/generic/488 | 76 +++
tests/generic/488.out | 7
tests/generic/group | 1 +
3 files change
w);
> +
> +static struct attribute *btrfs_send_attrs[] = {
> + BTRFS_ATTR_PTR(, stream_version),
> + NULL
> +};
> +
> +static const struct attribute_group btrfs_send_attr_group = {
> + .name = "send",
> + .attrs = btrfs_send_attrs,
> +};
>
On Tue, May 08, 2018 at 10:06:51PM -0400, Howard McLauchlan wrote:
> From: Howard McLauchlan
>
> Presently btrfs send/receive does not propagate inode attribute flags;
> all chattr operations are effectively discarded upon transmission.
>
> This patch adds kernel support for inode attribute flag
IMES, but it includes OTIME too. */
I've had feature requests for the ability to send compressed data
without decompressing it on the send side or the receive side. Could you
add a placeholder BTRFS_SEND_C_WRITE_COMPRESSED here? This way when I
get around to implementing it I won't have to do a v3.
Other than that,
Reviewed-by: Omar Sandoval
--
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 http://vger.kernel.org/majordomo-info.html
On Wed, May 16, 2018 at 09:48:58AM +0800, Eryu Guan wrote:
> On Wed, May 09, 2018 at 11:21:55PM -0700, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > Btrfs has a bug where we can prematurely ENOSPC if we have lots of
> > orphaned files, i.e., deleted files which
On Mon, May 14, 2018 at 09:40:19AM +0100, Dimitri John Ledkov wrote:
> Are both of these meant to be public libraries, installed on the user
> systems, and available in .so variant as well for 3rd party
> development and public dynamic linking?
>
> Or are these private internal libraries, which ar
On Fri, May 11, 2018 at 09:32:28PM +0100, Al Viro wrote:
> On Fri, May 11, 2018 at 01:30:01PM -0700, Omar Sandoval wrote:
> > On Fri, May 11, 2018 at 09:05:38PM +0100, Al Viro wrote:
> > > On Thu, May 10, 2018 at 11:30:10PM -0700, Omar Sandoval wrote:
> > > > do_b
On Fri, May 11, 2018 at 09:05:38PM +0100, Al Viro wrote:
> On Thu, May 10, 2018 at 11:30:10PM -0700, Omar Sandoval wrote:
> > do_blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
> > struct block_device *bdev, struct iov_iter *iter,
> >
From: Omar Sandoval
btrfs_orphan_commit_root() tries to delete an orphan item for a
subvolume in the tree root, but we don't actually insert that item in
the first place. See commit 0a0d4415e338 ("Btrfs: delete dead code in
btrfs_orphan_add()"). We can get rid of it.
Reviewed-
From: Omar Sandoval
Currently, we keep space reserved for all inode orphan items until the
inode is evicted (i.e., all references to it are dropped). We hit an
issue where an application would keep a bunch of deleted files open (by
design) and thus keep a large amount of space reserved, causing
From: Omar Sandoval
btrfs_link() calls btrfs_orphan_del() if it's linking an O_TMPFILE but
it doesn't reserve space to do so. Even before the removal of the
orphan_block_rsv it wasn't using it.
Fixes: ef3b9af50bfa ("Btrfs: implement inode_operations callback tmpfile")
R
From: Omar Sandoval
Now that we don't keep long-standing reservations for orphan items,
root->orphan_block_rsv isn't used. We can git rid of it, along with:
- root->orphan_lock, which was used to protect root->orphan_block_rsv
- root->orphan_inodes, which was used
From: Omar Sandoval
In btrfs_evict_inode(), if btrfs_truncate_inode_items() fails, the inode
item will still be in the tree but we still return the ino to the ino
cache. That will blow up later when someone tries to allocate that ino,
so don't return it to the cache.
Fixes: 581bb050941b (&
From: Omar Sandoval
Now that we don't add orphan items for truncate, there can't be races on
adding or deleting an orphan item, so this bit is unnecessary.
Reviewed-by: Nikolay Borisov
Signed-off-by: Omar Sandoval
---
fs/btrfs/btrfs_inode.h | 1 -
fs/btrfs/inode.c
From: Omar Sandoval
btrfs_free_extent() can fail because of ENOMEM. There's no reason to
panic here, we can just abort the transaction.
Fixes: f4b9aa8d3b87 ("btrfs_truncate")
Reviewed-by: Nikolay Borisov
Signed-off-by: Omar Sandoval
---
fs/btrfs/inode.c | 5 -
1
From: Omar Sandoval
Hi,
This is the fourth (and hopefully final) version of the orphan item
early ENOSPC and related fixes.
Changes since v3:
- Changed another stale comment in patch 1
- Moved BTRFS_INODE_ORPHAN_META_RESERVED flag removal to patch 10
instead of patch 9
- Moved inode runtime
From: Omar Sandoval
Commit a41ad394a03b ("Btrfs: convert to the new truncate sequence")
changed btrfs_setsize() to call truncate_setsize() instead of
vmtruncate() but didn't update the comment above it. truncate_setsize()
never fails (the IS_SWAPFILE() check happens elsewhere)
From: Omar Sandoval
We got rid of BTRFS_INODE_HAS_ORPHAN_ITEM and
BTRFS_INODE_ORPHAN_META_RESERVED, so we can renumber the flags to make
them consecutive again.
Signed-off-by: Omar Sandoval
---
fs/btrfs/btrfs_inode.h | 16
1 file changed, 8 insertions(+), 8 deletions(-)
diff
From: Omar Sandoval
The truncate loop in btrfs_evict_inode() does two things at once:
- It refills the temporary block reserve, potentially stealing from the
global reserve or committing
- It calls btrfs_truncate_inode_items()
The tangle of continues hides the fact that these two steps are
From: Omar Sandoval
btrfs_truncate_inode_items() uses two variables for error handling, ret
and err. These are not handled consistently, leading to a couple of
bugs.
- Errors from btrfs_del_items() are handled but not propagated to the
caller
- If btrfs_run_delayed_refs() fails and aborts the
From: Omar Sandoval
Currently, we insert an orphan item during a truncate so that if there's
a crash, we don't leak extents past the on-disk i_size. However, since
commit 7f4f6e0a3f6d ("Btrfs: only update disk_i_size as we remove
extents"), we keep disk_i_size in sync with th
On Fri, May 11, 2018 at 06:49:16PM +0200, David Sterba wrote:
> On Fri, May 11, 2018 at 05:25:50PM +0100, Filipe Manana wrote:
> > On Fri, May 11, 2018 at 4:57 PM, David Sterba wrote:
> > > The dedupe range is 16 MiB, with 4KiB pages and 8 byte pointers, the
> > > arrays can be 32KiB large. To avo
On Fri, May 11, 2018 at 12:53:36PM +0300, Nikolay Borisov wrote:
>
>
> On 11.05.2018 09:30, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > Hi, everyone,
> >
> > Btrfs currently abuses current->journal_info in btrfs_direct_IO() in
> > or
On Fri, May 11, 2018 at 12:19:43PM +0200, David Sterba wrote:
> On Fri, May 11, 2018 at 12:56:06AM -0700, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > Commit a41ad394a03b ("Btrfs: convert to the new truncate sequence")
> > changed vmtruncate() to tru
On Fri, May 11, 2018 at 06:51:30PM +0200, David Sterba wrote:
> On Fri, May 11, 2018 at 12:10:38PM -0400, Josef Bacik wrote:
> > I told him to do this, these flags aren't exposed anywhere are they?
> > They are in-kernel specific stuff, please tell me we aren't exposing
> > these via sysfs?
>
> No
From: Omar Sandoval
Currently, we keep space reserved for all inode orphan items until the
inode is evicted (i.e., all references to it are dropped). We hit an
issue where an application would keep a bunch of deleted files open (by
design) and thus keep a large amount of space reserved, causing
From: Omar Sandoval
Currently, we insert an orphan item during a truncate so that if there's
a crash, we don't leak extents past the on-disk i_size. However, since
commit 7f4f6e0a3f6d ("Btrfs: only update disk_i_size as we remove
extents"), we keep disk_i_size in sync with th
From: Omar Sandoval
btrfs_orphan_commit_root() tries to delete an orphan item for a
subvolume in the tree root, but we don't actually insert that item in
the first place. See commit 0a0d4415e338 ("Btrfs: delete dead code in
btrfs_orphan_add()"). We can get rid of it.
Sign
From: Omar Sandoval
Now that we don't keep long-standing reservations for orphan items,
root->orphan_block_rsv isn't used. We can git rid of it, along with
root->orphan_lock, which was used to protect it, root->orphan_inodes,
which was used as a refcount for it, and btrfs
From: Omar Sandoval
In btrfs_evict_inode(), if btrfs_truncate_inode_items() fails, the inode
item will still be in the tree but we still return the ino to the ino
cache. That will blow up later when someone tries to allocate that ino,
so don't return it to the cache.
Fixes: 581bb050941b (&
From: Omar Sandoval
The truncate loop in btrfs_evict_inode() does two things at once:
- It refills the temporary block reserve, potentially stealing from the
global reserve or committing
- It calls btrfs_truncate_inode_items()
The tangle of continues hides the fact that these two steps are
From: Omar Sandoval
btrfs_link() calls btrfs_orphan_del() if it's linking an O_TMPFILE but
it doesn't reserve space to do so. Even before the removal of the
orphan_block_rsv it wasn't using it.
Fixes: ef3b9af50bfa ("Btrfs: implement inode_operations callback tmpfile")
R
From: Omar Sandoval
Now that we don't add orphan items for truncate, there can't be races on
adding or deleting an orphan item, so this bit is unnecessary.
Signed-off-by: Omar Sandoval
---
fs/btrfs/btrfs_inode.h | 13
fs/btrfs/inode.c
From: Omar Sandoval
btrfs_free_extent() can fail because of ENOMEM. There's no reason to
panic here, we can just abort the transaction.
Fixes: f4b9aa8d3b87 ("btrfs_truncate")
Reviewed-by: Nikolay Borisov
Signed-off-by: Omar Sandoval
---
fs/btrfs/inode.c | 5 -
1
From: Omar Sandoval
btrfs_truncate_inode_items() uses two variables for error handling, ret
and err. These are not handled consistently, leading to a couple of
bugs.
- Errors from btrfs_del_items() are handled but not propagated to the
caller
- If btrfs_run_delayed_refs() fails and aborts the
From: Omar Sandoval
Commit a41ad394a03b ("Btrfs: convert to the new truncate sequence")
changed vmtruncate() to truncate_setsize() but didn't update the comment
above it. truncate_setsize() never fails (the IS_SWAPFILE() check
happens elsewhere), so remove the comment.
Sign
From: Omar Sandoval
Hi,
This is v3 of the fixes for the orphan item early ENOSPC issue we hit at
Facebook. The big change is that I now also got rid of
BTRFS_INODE_HAS_ORPHAN_ITEM (thanks, Nikolay) and shuffled the patches
around so there is less churn.
Changes since v2:
- Add patch 5 to get
On Thu, May 10, 2018 at 11:48:28PM -0700, Omar Sandoval wrote:
> On Fri, May 11, 2018 at 09:44:36AM +0300, Nikolay Borisov wrote:
> >
> >
> > On 11.05.2018 03:11, Omar Sandoval wrote:
> > > From: Omar Sandoval
> > >
> > > Now that we don&#x
On Fri, May 11, 2018 at 10:01:34AM +0300, Nikolay Borisov wrote:
>
>
> On 11.05.2018 03:11, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > btrfs_orphan_commit_root() tries to delete an orphan item for a
> > subvolume in the tree root, but we don'
On Fri, May 11, 2018 at 09:38:15AM +0300, Nikolay Borisov wrote:
>
>
> On 11.05.2018 03:11, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > Currently, we keep space reserved for all inode orphan items until the
> > inode is evicted (i.e., all referen
On Fri, May 11, 2018 at 09:44:36AM +0300, Nikolay Borisov wrote:
>
>
> On 11.05.2018 03:11, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > Now that we don't keep long-standing reservations for orphan items,
> > root->orphan_block_rsv isn&
From: Omar Sandoval
Btrfs abuses current->journal_info in btrfs_direct_IO() in order to pass
around some state to get_block() and submit_io(). The generic DIO code
already provides bh_result->b_private as a way to pass data between
calls to get_block() and end_io(), but it is always initi
From: Omar Sandoval
Hi, everyone,
Btrfs currently abuses current->journal_info in btrfs_direct_IO() in
order to pass around some state to get_block() and submit_io(). This
hack is ugly and unnecessary because the data we pass around is only
used in one call frame. Robbie Ko also pointed out
From: Omar Sandoval
Now that we can pass around the struct btrfs_dio_data through the
different callbacks generically, we don't need to shove it in
current->journal_info.
Signed-off-by: Omar Sandoval
---
fs/btrfs/inode.c | 33 +++--
1 file changed, 7 in
From: Omar Sandoval
get_block() already has access to this through bh_result->b_private, and
it gets passed to end_io() as private, so do the same for submit_io().
Along with the previous change, this will allow us to get rid Btrfs'
current->journal_info abuse in btrfs_direct_IO().
From: Omar Sandoval
btrfs_truncate_inode_items() uses two variables for error handling, ret
and err. These are not handled consistently, leading to a couple of
bugs.
- Errors from btrfs_del_items() are handled but not propagated to the
caller
- If btrfs_run_delayed_refs() fails and aborts the
From: Omar Sandoval
Currently, we insert an orphan item during a truncate so that if there's
a crash, we don't leak extents past the on-disk i_size. However, since
commit 7f4f6e0a3f6d ("Btrfs: only update disk_i_size as we remove
extents"), we keep disk_i_size in sync with th
From: Omar Sandoval
In btrfs_evict_inode(), if btrfs_truncate_inode_items() fails, the inode
item will still be in the tree but we still return the ino to the ino
cache. That will blow up later when someone tries to allocate that ino,
so don't return it to the cache.
Fixes: 581bb050941b (&
From: Omar Sandoval
Now that we don't keep long-standing reservations for orphan items,
root->orphan_block_rsv isn't used. We can git rid of it and
root->orphan_lock, which was used to protect it.
Signed-off-by: Omar Sandoval
---
fs/btrfs/ctree.h | 5 -
fs/btrfs/di
From: Omar Sandoval
We have the same error handling code duplicated all over the place. Put
it all in one place, and while we're here, get rid of the weird
btrfs_orphan_del() trans == NULL case and just clear the orphan item bit
directly since that's all it does anymore.
Signed-of
From: Omar Sandoval
btrfs_orphan_commit_root() tries to delete an orphan item for a
subvolume in the tree root, but we don't actually insert that item in
the first place. See commit 0a0d4415e338 ("Btrfs: delete dead code in
btrfs_orphan_add()"). We can get rid of it.
root->
From: Omar Sandoval
Currently, if btrfs_insert_orphan_item() fails, we release the reserved
space and decrement root->orphan_inodes. However, we also ignore -EEXIST
errors, so we still want the space reservation for when we delete the
item, and we still need to count the orphan because we
From: Omar Sandoval
btrfs_link() calls btrfs_orphan_del() if it's linking an O_TMPFILE but
it doesn't reserve space to do so. Even before the removal of the
orphan_block_rsv it wasn't using it.
Fixes: ef3b9af50bfa ("Btrfs: implement inode_operations callback tmpfile")
R
From: Omar Sandoval
Currently, we keep space reserved for all inode orphan items until the
inode is evicted (i.e., all references to it are dropped). We hit an
issue where an application would keep a bunch of deleted files open (by
design) and thus keep a large amount of space reserved, causing
From: Omar Sandoval
Hi,
This is v2 of the fixes for the orphan item early ENOSPC issue we hit at
Facebook.
Changes since v1:
- Added two extra cleanups, patches 10 and 11
- Added a forgotten clear of the orphan bit in patch 8
- Reworded titles of patches 6 and 9
- Added people's reviewe
From: Omar Sandoval
btrfs_free_extent() can fail because of ENOMEM. There's no reason to
panic here, we can just abort the transaction.
Fixes: f4b9aa8d3b87 ("btrfs_truncate")
Reviewed-by: Nikolay Borisov
Signed-off-by: Omar Sandoval
---
fs/btrfs/inode.c | 5 -
1
From: Omar Sandoval
Commit a41ad394a03b ("Btrfs: convert to the new truncate sequence")
changed vmtruncate() to truncate_setsize() but didn't update the comment
above it. truncate_setsize() never fails (the IS_SWAPFILE() check
happens elsewhere), so remove the comment.
Sign
From: Omar Sandoval
The truncate loop in btrfs_evict_inode() does two things at once:
- It refills the temporary block reserve, potentially stealing from the
global reserve or committing
- It calls btrfs_truncate_inode_items()
The tangle of continues hides the fact that these two steps are
On Thu, May 10, 2018 at 11:50:11AM +0300, Nikolay Borisov wrote:
>
>
> On 10.05.2018 09:21, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > Currently, we keep space reserved for all inode orphan items until the
> > inode is evicted (i.e., all referen
On Thu, May 10, 2018 at 11:29:18AM +0300, Nikolay Borisov wrote:
>
>
> On 10.05.2018 09:21, Omar Sandoval wrote:
> > From: Omar Sandoval
> >
> > In btrfs_evict_inode(), if btrfs_truncate_inode_items() fails, the inode
> > item will still be in the tree but w
From: Omar Sandoval
The truncate loop in btrfs_evict_inode() does two things at once:
- It refills the temporary block reserve, potentially stealing from the
global reserve or committing
- It calls btrfs_truncate_inode_items()
The tangle of continues hides the fact that these two steps are
From: Omar Sandoval
Btrfs has a bug where we can prematurely ENOSPC if we have lots of
orphaned files, i.e., deleted files which are still open. Add a test
which repeatedly creates and deletes a file while keeping all of the
file descriptors open. This should succeed but doesn't on Btrfs wi
From: Omar Sandoval
Currently, we insert an orphan item during a truncate so that if there's
a crash, we don't leak extents past the on-disk i_size. However, since
commit 7f4f6e0a3f6d ("Btrfs: only update disk_i_size as we remove
extents"), we keep disk_i_size in sync with th
From: Omar Sandoval
Currently, if btrfs_insert_orphan_item() fails, we release the reserved
space and decrement root->orphan_inodes. However, we also ignore -EEXIST
errors, so we still want the space reservation for when we delete the
item, and we still need to count the orphan because we
From: Omar Sandoval
btrfs_free_extent() can fail because of ENOMEM. There's no reason to
panic here, we can just abort the transaction.
Fixes: f4b9aa8d3b87 ("btrfs_truncate")
Signed-off-by: Omar Sandoval
---
fs/btrfs/inode.c | 5 -
1 file changed, 4 insertions(+), 1 de
From: Omar Sandoval
In btrfs_evict_inode(), if btrfs_truncate_inode_items() fails, the inode
item will still be in the tree but we still return the ino to the ino
cache. That will blow up later when someone tries to allocate that ino,
so don't return it to the cache.
Fixes: 581bb050941b (&
From: Omar Sandoval
Now that we don't keep long-standing reservations for orphan items, we
can get rid of root->orphan_lock and root->orphan_block_rsv, as well as
some related functions.
Signed-off-by: Omar Sandoval
---
fs/btrfs/ctree.h | 5 -
fs/btrfs/disk-io
From: Omar Sandoval
btrfs_link() calls btrfs_orphan_del() if it's linking an O_TMPFILE but
it doesn't reserve space to do so. Even before the removal of the
orphan_block_rsv it wasn't using it.
Fixes: ef3b9af50bfa ("Btrfs: implement inode_operations callback tmpfile&q
From: Omar Sandoval
Currently, we keep space reserved for all inode orphan items until the
inode is evicted (i.e., all references to it are dropped). We hit an
issue where an application would keep a bunch of deleted files open (by
design) and thus keep a large amount of space reserved, causing
From: Omar Sandoval
btrfs_truncate_inode_items() uses two variables for error handling, ret
and err. These are not handled consistently, leading to a couple of
bugs.
- Errors from btrfs_del_items() are handled but not propagated to the
caller
- If btrfs_run_delayed_refs() fails and aborts the
From: Omar Sandoval
Commit a41ad394a03b ("Btrfs: convert to the new truncate sequence")
changed vmtruncate() to truncate_setsize() but didn't update the comment
above it. truncate_setsize() never fails (the IS_SWAPFILE() check
happens elsewhere), so remove the comment.
Sign
From: Omar Sandoval
Hi,
At Facebook we hit an early ENOSPC issue which we tracked down to the
reservations for orphan items of deleted-but-still-open files. The
primary function of this series is to fix that bug, but I ended up
uncovering a pile of other issues in the process, most notably that
On Wed, May 09, 2018 at 06:35:25PM +0800, robbieko wrote:
> From: Robbie Ko
>
> When send process requires memory allocation, shrinker may be triggered
> due to insufficient memory.
> Then evict_inode gets called when inode is freed, and this function
> may need to start transaction.
> However, t
From: Omar Sandoval
generic/427 creates a 256 MB filesystem and then writes a 200 MB file,
which fails on Btrfs if mixed mode is not enabled. Raise the threshold
to 1GB, which is where we typically recommend mixed mode.
Signed-off-by: Omar Sandoval
---
common/rc | 2 +-
1 file changed, 1
From: Omar Sandoval
generic/427 fails on Btrfs with a cryptic "pread: Success" message. This
is because an aio pwrite fails with ENOSPC, so the file isn't as long as
we expect it to be. Make sure we check the result of the aio writes and
also print a more explicit message for sho
401 - 500 of 1059 matches
Mail list logo
