platform/kernel/kernel-mfld-blackbay.git
13 years agoBtrfs: Fix page leak in compressed writeback path
Yan, Zheng [Mon, 6 Dec 2010 07:02:36 +0000 (07:02 +0000)]
Btrfs: Fix page leak in compressed writeback path

"start + num_bytes >= actual_end" can happen when compressed page writeback races
with file truncation. In that case we need unlock and release pages past the end
of file.

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: do not BUG if we fail to remove the orphan item for dead snapshots
Josef Bacik [Wed, 8 Dec 2010 17:24:01 +0000 (12:24 -0500)]
Btrfs: do not BUG if we fail to remove the orphan item for dead snapshots

Not being able to delete an orphan item isn't a horrible thing.  The worst that
happens is the next time around we try and do the orphan cleanup and we can't
find the referenced object and just delete the item and move on.

Signed-off-by: Josef Bacik <josef@redhat.com>
13 years agoBtrfs: fixup return code for btrfs_del_orphan_item
Josef Bacik [Wed, 8 Dec 2010 17:22:34 +0000 (12:22 -0500)]
Btrfs: fixup return code for btrfs_del_orphan_item

If the orphan item doesn't exist, we return 1, which doesn't make any sense to
the callers.  Instead return -ENOENT if we didn't find the item.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
13 years agoBtrfs: do not do fast caching if we are allocating blocks for tree_root
Josef Bacik [Wed, 8 Dec 2010 14:15:11 +0000 (09:15 -0500)]
Btrfs: do not do fast caching if we are allocating blocks for tree_root

Since the fast caching uses normal tree locking, we can possibly deadlock if we
get to the caching via a btrfs_search_slot() on the tree_root.  So just check to
see if the root we are on is the tree root, and just don't do the fast caching.

Reported-by: Sage Weil <sage@newdream.net>
Signed-off-by: Josef Bacik <josef@redhat.com>
13 years agoBtrfs: deal with space cache errors better
Josef Bacik [Fri, 3 Dec 2010 18:17:53 +0000 (13:17 -0500)]
Btrfs: deal with space cache errors better

Currently if the space cache inode generation number doesn't match the
generation number in the space cache header we will just fail to load the space
cache, but we won't mark the space cache as an error, so we'll keep getting that
error each time somebody tries to cache that block group until we actually clear
the thing.  Fix this by marking the space cache as having an error so we only
get the message once.  This patch also makes it so that we don't try and setup
space cache for a block group that isn't cached, since we won't be able to write
it out anyway.  None of these problems are actual problems, they are just
annoying and sub-optimal.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
13 years agoBtrfs: fix use after free in O_DIRECT
Josef Bacik [Fri, 19 Nov 2010 14:41:10 +0000 (09:41 -0500)]
Btrfs: fix use after free in O_DIRECT

This fixes a bug where we use dip after we have freed it.  Instead just use the
file_offset that was passed to the function.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
13 years agoBtrfs: don't use migrate page without CONFIG_MIGRATION
Chris Mason [Mon, 29 Nov 2010 14:49:11 +0000 (09:49 -0500)]
Btrfs: don't use migrate page without CONFIG_MIGRATION

Fixes compile error

Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: deal with DIO bios that span more than one ordered extent
Chris Mason [Mon, 29 Nov 2010 00:56:33 +0000 (19:56 -0500)]
Btrfs: deal with DIO bios that span more than one ordered extent

The new DIO bio splitting code has problems when the bio
spans more than one ordered extent.  This will happen as the
generic DIO code merges our get_blocks calls together into
a bigger single bio.

This fixes things by walking forward in the ordered extent
code finding all the overlapping ordered extents and completing them
all at once.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: setup blank root and fs_info for mount time
Josef Bacik [Fri, 19 Nov 2010 19:59:15 +0000 (14:59 -0500)]
Btrfs: setup blank root and fs_info for mount time

There is a problem with how we use sget, it searches through the list of supers
attached to the fs_type looking for a super with the same fs_devices as what
we're trying to mount.  This depends on sb->s_fs_info being filled, but we don't
fill that in until we get to btrfs_fill_super, so we could hit supers on the
fs_type super list that have a null s_fs_info.  In order to fix that we need to
go ahead and setup a blank root with a blank fs_info to hold fs_devices, that
way our test will work out right and then we can set s_fs_info in
btrfs_set_super, and then open_ctree will simply use our pre-allocated root and
fs_info when setting everything up.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: fix fiemap
Josef Bacik [Tue, 23 Nov 2010 19:36:57 +0000 (19:36 +0000)]
Btrfs: fix fiemap

There are two big problems currently with FIEMAP

1) We return extents for holes.  This isn't supposed to happen, we just don't
return extents for holes and then userspace interprets the lack of an extent as
a hole.

2) We sometimes don't set FIEMAP_EXTENT_LAST properly.  This is because we wait
to see a EXTENT_FLAG_VACANCY flag on the em, but this won't happen if say we ask
fiemap to map up to the last extent in a file, and there is nothing but holes up
to the i_size.  To fix this we need to lookup the last extent in this file and
save the logical offset, so if we happen to try and map that extent we can be
sure to set FIEMAP_EXTENT_LAST.

With this patch we now pass xfstest 225, which we never have before.

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs - fix race between btrfs_get_sb() and umount
Ian Kent [Mon, 22 Nov 2010 02:21:38 +0000 (02:21 +0000)]
Btrfs - fix race between btrfs_get_sb() and umount

When mounting a btrfs file system btrfs_test_super() may attempt to
use sb->s_fs_info, the btrfs root, of a super block that is going away
and that has had the btrfs root set to NULL in its ->put_super(). But
if the super block is going away it cannot be an existing super block
so we can return false in this case.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: update inode ctime when using links
Josef Bacik [Tue, 23 Nov 2010 19:50:59 +0000 (19:50 +0000)]
Btrfs: update inode ctime when using links

Currently we fail xfstest 236 because we're not updating the inode ctime on
link.  This is a simple fix, and makes it so we pass 236 now.

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: make sure new inode size is ok in fallocate
Josef Bacik [Mon, 22 Nov 2010 18:55:39 +0000 (18:55 +0000)]
Btrfs: make sure new inode size is ok in fallocate

We have been failing xfstest 228 forever, because we don't check to make sure
the new inode size is acceptable as far as RLIMIT is concerned.  Just check to
make sure it's ok to create a inode with this new size and error out if not.
With this patch we now pass 228.

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: fix typo in fallocate to make it honor actual size
Josef Bacik [Mon, 22 Nov 2010 18:50:32 +0000 (18:50 +0000)]
Btrfs: fix typo in fallocate to make it honor actual size

There is a typo in __btrfs_prealloc_file_range() where we set the i_size to
actual_len/cur_offset, and then just set it to cur_offset again, and do the same
with btrfs_ordered_update_i_size().  This fixes it back to keeping i_size in a
local variable and then updating i_size properly.  Tested this with

xfs_io -F -f -c "falloc 0 1" -c "pwrite 0 1" foo

stat'ing foo gives us a size of 1 instead of 4096 like it was.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: avoid NULL pointer deref in try_release_extent_buffer
Chris Mason [Mon, 22 Nov 2010 03:27:44 +0000 (22:27 -0500)]
Btrfs: avoid NULL pointer deref in try_release_extent_buffer

If we fail to find a pointer in the radix tree, don't try
to deref the NULL one we do have.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: make btrfs_add_nondir take parent inode as an argument
Josef Bacik [Fri, 19 Nov 2010 20:36:11 +0000 (20:36 +0000)]
Btrfs: make btrfs_add_nondir take parent inode as an argument

Everybody who calls btrfs_add_nondir just passes in the dentry of the new file
and then dereference dentry->d_parent->d_inode, but everybody who calls
btrfs_add_nondir() are already passed the parent's inode.  So instead of
dereferencing dentry->d_parent, just make btrfs_add_nondir take the dir inode as
an argument and pass that along so we don't have to worry about d_parent.
Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: hold i_mutex when calling btrfs_log_dentry_safe
Josef Bacik [Fri, 19 Nov 2010 20:36:10 +0000 (20:36 +0000)]
Btrfs: hold i_mutex when calling btrfs_log_dentry_safe

Since we walk up the path logging all of the parts of the inode's path, we need
to hold i_mutex to make sure that the inode is not renamed while we're logging
everything.  btrfs_log_dentry_safe does dget_parent and all of that jazz, but we
may get unexpected results if the rename changes the inode's location while
we're higher up the path logging those dentries, so do this for safety reasons.
Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: use dget_parent where we can UPDATED
Josef Bacik [Sat, 20 Nov 2010 09:48:00 +0000 (09:48 +0000)]
Btrfs: use dget_parent where we can UPDATED

There are lots of places where we do dentry->d_parent->d_inode without holding
the dentry->d_lock.  This could cause problems with rename.  So instead we need
to use dget_parent() and hold the reference to the parent as long as we are
going to use it's inode and then dput it at the end.

Signed-off-by: Josef Bacik <josef@redhat.com>
Cc: raven@themaw.net
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: fix more ESTALE problems with NFS
Josef Bacik [Fri, 19 Nov 2010 02:18:02 +0000 (02:18 +0000)]
Btrfs: fix more ESTALE problems with NFS

When creating new inodes we don't setup inode->i_generation.  So if we generate
an fh with a newly created inode we save the generation of 0, but if we flush
the inode to disk and have to read it back when getting the inode on the server
we'll have the right i_generation, so gens wont match and we get ESTALE.  This
patch properly sets inode->i_generation when we create the new inode and now I'm
no longer getting ESTALE.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: handle NFS lookups properly
Josef Bacik [Wed, 17 Nov 2010 18:54:54 +0000 (18:54 +0000)]
Btrfs: handle NFS lookups properly

People kept reporting NFS issues, specifically getting ESTALE alot.  I figured
out how to reproduce the problem

SERVER
mkfs.btrfs /dev/sda1
mount /dev/sda1 /mnt/btrfs-test
<add /mnt/btrfs-test to /etc/exports>
btrfs subvol create /mnt/btrfs-test/foo
service nfs start

CLIENT
mount server:/mnt/btrfs /mnt/test
cd /mnt/test/foo
ls

SERVER
echo 3 > /proc/sys/vm/drop_caches

CLIENT
ls <-- get an ESTALE here

This is because the standard way to lookup a name in nfsd is to use readdir, and
what it does is do a readdir on the parent directory looking for the inode of
the child.  So in this case the parent being / and the child being foo.  Well
subvols all have the same inode number, so doing a readdir of / looking for
inode 256 will return '.', which obviously doesn't match foo.  So instead we
need to have our own .get_name so that we can find the right name.

Our .get_name will either lookup the inode backref or the root backref,
whichever we're looking for, and return the name we find.  Running the above
reproducer with this patch results in everything acting the way its supposed to.
Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agobtrfs: make 1-bit signed fileds unsigned
Mariusz Kozlowski [Sat, 20 Nov 2010 12:03:07 +0000 (12:03 +0000)]
btrfs: make 1-bit signed fileds unsigned

Fixes these sparse warnings:
fs/btrfs/ctree.h:811:17: error: dubious one-bit signed bitfield
fs/btrfs/ctree.h:812:20: error: dubious one-bit signed bitfield
fs/btrfs/ctree.h:813:19: error: dubious one-bit signed bitfield

Signed-off-by: Mariusz Kozlowski <mk@lab.zgora.pl>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agobtrfs: Show device attr correctly for symlinks
Li Zefan [Fri, 19 Nov 2010 02:05:24 +0000 (02:05 +0000)]
btrfs: Show device attr correctly for symlinks

Symlinks and files of other types show different device numbers, though
they are on the same partition:

 $ touch tmp; ln -s tmp tmp2; stat tmp tmp2
   File: `tmp'
   Size: 0          Blocks: 0          IO Block: 4096   regular empty file
 Device: 15h/21d Inode: 984027      Links: 1
 --- snip ---
   File: `tmp2' -> `tmp'
   Size: 3          Blocks: 0          IO Block: 4096   symbolic link
 Device: 13h/19d Inode: 984028      Links: 1

Reported-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agobtrfs: Set file size correctly in file clone
Li Zefan [Fri, 19 Nov 2010 01:36:34 +0000 (01:36 +0000)]
btrfs: Set file size correctly in file clone

Set src_offset = 0, src_length = 20K, dest_offset = 20K. And the
original filesize of the dest file 'file2' is 30K:

  # ls -l /mnt/file2
  -rw-r--r-- 1 root root 30720 Nov 18 16:42 /mnt/file2

Now clone file1 to file2, the dest file should be 40K, but it
still shows 30K:

  # ls -l /mnt/file2
  -rw-r--r-- 1 root root 30720 Nov 18 16:42 /mnt/file2

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agobtrfs: Check if dest_offset is block-size aligned before cloning file
Li Zefan [Fri, 19 Nov 2010 01:36:10 +0000 (01:36 +0000)]
btrfs: Check if dest_offset is block-size aligned before cloning file

We've done the check for src_offset and src_length, and We should
also check dest_offset, otherwise we'll corrupt the destination
file:

  (After cloning file1 to file2 with unaligned dest_offset)
  # cat /mnt/file2
  cat: /mnt/file2: Input/output error

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: handle the space_cache option properly
Josef Bacik [Fri, 19 Nov 2010 13:40:41 +0000 (13:40 +0000)]
Btrfs: handle the space_cache option properly

When I added the clear_cache option I screwed up and took the break out of
the space_cache case statement, so whenever you mount with space_cache you also
get clear_cache, which does you no good if you say set space_cache in fstab so
it always gets set.  This patch adds the break back in properly.

Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agobtrfs: Fix early enospc because 'unused' calculated with wrong sign.
Arne Jansen [Fri, 12 Nov 2010 23:17:56 +0000 (23:17 +0000)]
btrfs: Fix early enospc because 'unused' calculated with wrong sign.

'unused' calculated with wrong sign in reserve_metadata_bytes().
This might have lead to unwanted over-reservations.

Signed-off-by: Arne Jansen <sensille@gmx.net>
Reviewed-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agobtrfs: fix panic caused by direct IO
Miao Xie [Mon, 22 Nov 2010 03:04:43 +0000 (03:04 +0000)]
btrfs: fix panic caused by direct IO

btrfs paniced when we write >64KB data by direct IO at one time.

Reproduce steps:
 # mkfs.btrfs /dev/sda5 /dev/sda6
 # mount /dev/sda5 /mnt
 # dd if=/dev/zero of=/mnt/tmpfile bs=100K count=1 oflag=direct

Then btrfs paniced:
mapping failed logical 1103155200 bio len 69632 len 12288
------------[ cut here ]------------
kernel BUG at fs/btrfs/volumes.c:3010!
[SNIP]
Pid: 1992, comm: btrfs-worker-0 Not tainted 2.6.37-rc1 #1 D2399/PRIMERGY
RIP: 0010:[<ffffffffa03d1462>]  [<ffffffffa03d1462>] btrfs_map_bio+0x202/0x210 [btrfs]
[SNIP]
Call Trace:
 [<ffffffffa03ab3eb>] __btrfs_submit_bio_done+0x1b/0x20 [btrfs]
 [<ffffffffa03a35ff>] run_one_async_done+0x9f/0xb0 [btrfs]
 [<ffffffffa03d3d20>] run_ordered_completions+0x80/0xc0 [btrfs]
 [<ffffffffa03d45a4>] worker_loop+0x154/0x5f0 [btrfs]
 [<ffffffffa03d4450>] ? worker_loop+0x0/0x5f0 [btrfs]
 [<ffffffffa03d4450>] ? worker_loop+0x0/0x5f0 [btrfs]
 [<ffffffff81083216>] kthread+0x96/0xa0
 [<ffffffff8100cec4>] kernel_thread_helper+0x4/0x10
 [<ffffffff81083180>] ? kthread+0x0/0xa0
 [<ffffffff8100cec0>] ? kernel_thread_helper+0x0/0x10

We fix this problem by splitting bios when we submit bios.

Reported-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Tested-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agobtrfs: cleanup duplicate bio allocating functions
Miao Xie [Mon, 22 Nov 2010 03:02:55 +0000 (03:02 +0000)]
btrfs: cleanup duplicate bio allocating functions

extent_bio_alloc() and compressed_bio_alloc() are similar, cleanup
similar source code.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agobtrfs: fix free dip and dip->csums twice
Miao Xie [Mon, 22 Nov 2010 03:01:39 +0000 (03:01 +0000)]
btrfs: fix free dip and dip->csums twice

bio_endio() will free dip and dip->csums, so dip and dip->csums twice will
be freed twice. Fix it.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
13 years agoBtrfs: add migrate page for metadata inode
Chris Mason [Mon, 22 Nov 2010 03:20:49 +0000 (22:20 -0500)]
Btrfs: add migrate page for metadata inode

Migrate page will directly call the btrfs btree writepage function,
which isn't actually allowed.

Our writepage assumes that you have locked the extent_buffer and
flagged the block as written.  Without doing these steps, we can
corrupt metadata blocks.

A later commit will remove the btree writepage function since
it is really only safely used internally by btrfs.  We
use writepages for everything else.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: deal with errors from updating the tree log
Chris Mason [Sat, 30 Oct 2010 11:34:24 +0000 (07:34 -0400)]
Btrfs: deal with errors from updating the tree log

During unlink we remove any references to the inode from
the tree log.  It can return -ENOENT and other errors,
and this changes the unlink code to deal with it.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: allow subvol deletion by unprivileged user with -o user_subvol_rm_allowed
Sage Weil [Fri, 29 Oct 2010 19:46:43 +0000 (15:46 -0400)]
Btrfs: allow subvol deletion by unprivileged user with -o user_subvol_rm_allowed

Add a mount option user_subvol_rm_allowed that allows users to delete a
(potentially non-empty!) subvol when they would otherwise we allowed to do
an rmdir(2).  We duplicate the may_delete() checks from the core VFS code
to implement identical security checks (minus the directory size check).
We additionally require that the user has write+exec permission on the
subvol root inode.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: make SNAP_DESTROY async
Sage Weil [Fri, 29 Oct 2010 19:41:32 +0000 (15:41 -0400)]
Btrfs: make SNAP_DESTROY async

There is no reason to force an immediate commit when deleting a snapshot.
Users have some expectation that space from a deleted snapshot be freed
immediately, but even if we do commit the reclaim is a background process.

If users _do_ want the deletion to be durable, they can call 'sync'.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: add SNAP_CREATE_ASYNC ioctl
Sage Weil [Fri, 29 Oct 2010 19:41:32 +0000 (15:41 -0400)]
Btrfs: add SNAP_CREATE_ASYNC ioctl

Create a snap without waiting for it to commit to disk.  The ioctl is
ordered such that subsequent operations will not be contained by the
created snapshot, and the commit is initiated, but the ioctl does not
wait for the snapshot to commit to disk.

We return the specific transid to userspace so that an application can wait
for this specific snapshot creation to commit via the WAIT_SYNC ioctl.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: add START_SYNC, WAIT_SYNC ioctls
Sage Weil [Fri, 29 Oct 2010 19:41:32 +0000 (15:41 -0400)]
Btrfs: add START_SYNC, WAIT_SYNC ioctls

START_SYNC will start a sync/commit, but not wait for it to
complete.  Any modification started after the ioctl returns is
guaranteed not to be included in the commit.  If a non-NULL
pointer is passed, the transaction id will be returned to
userspace.

WAIT_SYNC will wait for any in-progress commit to complete.  If a
transaction id is specified, the ioctl will block and then
return (success) when the specified transaction has committed.
If it has already committed when we call the ioctl, it returns
immediately.  If the specified transaction doesn't exist, it
returns EINVAL.

If no transaction id is specified, WAIT_SYNC will wait for the
currently committing transaction to finish it's commit to disk.
If there is no currently committing transaction, it returns
success.

These ioctls are useful for applications which want to impose an
ordering on when fs modifications reach disk, but do not want to
wait for the full (slow) commit process to do so.

Picky callers can take the transid returned by START_SYNC and
feed it to WAIT_SYNC, and be certain to wait only as long as
necessary for the transaction _they_ started to reach disk.

Sloppy callers can START_SYNC and WAIT_SYNC without a transid,
and provided they didn't wait too long between the calls, they
will get the same result.  However, if a second commit starts
before they call WAIT_SYNC, they may end up waiting longer for
it to commit as well.  Even so, a START_SYNC+WAIT_SYNC still
guarantees that any operation completed before the START_SYNC
reaches disk.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: async transaction commit
Sage Weil [Fri, 29 Oct 2010 19:37:34 +0000 (15:37 -0400)]
Btrfs: async transaction commit

Add support for an async transaction commit that is ordered such that any
subsequent operations will join the following transaction, but does not
wait until the current commit is fully on disk.  This avoids much of the
latency associated with the btrfs_commit_transaction for callers concerned
with serialization and not safety.

The wait_for_unblock flag controls whether we wait for the 'middle' portion
of commit_transaction to complete, which is necessary if the caller expects
some of the modifications contained in the commit to be available (this is
the case for subvol/snapshot creation).

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: fix deadlock in btrfs_commit_transaction
Sage Weil [Fri, 29 Oct 2010 19:37:34 +0000 (15:37 -0400)]
Btrfs: fix deadlock in btrfs_commit_transaction

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 deciding how long to wait when we wait.  Include a smp_mb()
before checking if the waitqueue is active to ensure the num_writers
is visible.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: fix lockdep warning on clone ioctl
Sage Weil [Fri, 29 Oct 2010 19:37:33 +0000 (15:37 -0400)]
Btrfs: fix lockdep warning on clone ioctl

I'm no lockdep expert, but this appears to make the lockdep warning go
away for the i_mutex locking in the clone ioctl.

Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: fix clone ioctl where range is adjacent to extent
Sage Weil [Fri, 29 Oct 2010 19:37:33 +0000 (15:37 -0400)]
Btrfs: fix clone ioctl where range is adjacent to extent

We had an edge case issue where the requested range was just
following an existing extent. Instead of skipping to the next
extent, we used the previous one which lead to having zero
sized extents.

Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: fix delalloc checks in clone ioctl
Sage Weil [Fri, 29 Oct 2010 19:37:33 +0000 (15:37 -0400)]
Btrfs: fix delalloc checks in clone ioctl

The lookup_first_ordered_extent() was done on the wrong inode, and the
->delalloc_bytes test was wrong, as the following
btrfs_wait_ordered_range() would only invoke a range write and wouldn't
write the entire file data range. Also, a bad parameter was passed to
btrfs_wait_ordered_range().

Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: drop unused variable in block_alloc_rsv
Chris Mason [Fri, 29 Oct 2010 19:17:41 +0000 (15:17 -0400)]
Btrfs: drop unused variable in block_alloc_rsv

The alloc_target variable is not really used.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: cleanup warnings from gcc 4.6 (nonbugs)
Andi Kleen [Fri, 29 Oct 2010 19:14:37 +0000 (15:14 -0400)]
Btrfs: cleanup warnings from gcc 4.6 (nonbugs)

These are all the cases where a variable is set, but not read which are
not bugs as far as I can see, but simply leftovers.

Still needs more review.

Found by gcc 4.6's new warnings

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Cc: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: Fix variables set but not read (bugs found by gcc 4.6)
Andi Kleen [Fri, 29 Oct 2010 19:14:31 +0000 (15:14 -0400)]
Btrfs: Fix variables set but not read (bugs found by gcc 4.6)

These are all the cases where a variable is set, but not
read which are really bugs.

- Couple of incorrect error handling fixed.
- One incorrect use of a allocation policy
- Some other things

Still needs more review.

Found by gcc 4.6's new warnings.

[akpm@linux-foundation.org: fix build.  Might have been bitrot]
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Cc: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: Use ERR_CAST helpers
Julia Lawall [Fri, 29 Oct 2010 19:14:23 +0000 (15:14 -0400)]
Btrfs: Use ERR_CAST helpers

Use ERR_CAST(x) rather than ERR_PTR(PTR_ERR(x)).  The former makes more
clear what is the purpose of the operation, which otherwise looks like a
no-op.

The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
type T;
T x;
identifier f;
@@

T f (...) { <+...
- ERR_PTR(PTR_ERR(x))
+ x
 ...+> }

@@
expression x;
@@

- ERR_PTR(PTR_ERR(x))
+ ERR_CAST(x)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Cc: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: use memdup_user helpers
Julia Lawall [Fri, 29 Oct 2010 19:14:18 +0000 (15:14 -0400)]
Btrfs: use memdup_user helpers

Use memdup_user when user data is immediately copied into the
allocated region.

The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression from,to,size,flag;
position p;
identifier l1,l2;
@@

-  to = \(kmalloc@p\|kzalloc@p\)(size,flag);
+  to = memdup_user(from,size);
   if (
-      to==NULL
+      IS_ERR(to)
                 || ...) {
   <+... when != goto l1;
-  -ENOMEM
+  PTR_ERR(to)
   ...+>
   }
-  if (copy_from_user(to, from, size) != 0) {
-    <+... when != goto l2;
-    -EFAULT
-    ...+>
-  }
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Cc: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: fix raid code for removing missing drives
Chris Mason [Thu, 28 Oct 2010 19:30:42 +0000 (15:30 -0400)]
Btrfs: fix raid code for removing missing drives

When btrfs is mounted in degraded mode, it has some internal structures
to track the missing devices.  This missing device is setup as readonly,
but the mapping code can get upset when we try to write to it.

This changes the mapping code to return -EIO instead of oops when we try
to write to the readonly device.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: Switch the extent buffer rbtree into a radix tree
Miao Xie [Wed, 27 Oct 2010 00:57:29 +0000 (20:57 -0400)]
Btrfs: Switch the extent buffer rbtree into a radix tree

This patch reduces the CPU time spent in the extent buffer search by using the
radix tree instead of the rbtree and using the rcu lock instead of the spin
lock.

I did a quick test by the benchmark tool[1] and found the patch improve the
file creation/deletion performance problem that I have reported[2].

Before applying this patch:
Create files:
Total files: 50000
Total time: 0.971531
Average time: 0.000019
Delete files:
Total files: 50000
Total time: 1.366761
Average time: 0.000027

After applying this patch:
Create files:
Total files: 50000
Total time: 0.927455
Average time: 0.000019
Delete files:
Total files: 50000
Total time: 1.292280
Average time: 0.000026

[1] http://marc.info/?l=linux-btrfs&m=128212635122920&q=p3
[2] http://marc.info/?l=linux-btrfs&m=128212635122920&w=2

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: restructure try_release_extent_buffer()
Miao Xie [Wed, 27 Oct 2010 00:57:29 +0000 (20:57 -0400)]
Btrfs: restructure try_release_extent_buffer()

restructure try_release_extent_buffer() and write a function to release the
extent buffer. It will be used later.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: use the flusher threads for delalloc throttling
Chris Mason [Tue, 26 Oct 2010 17:40:45 +0000 (13:40 -0400)]
Btrfs: use the flusher threads for delalloc throttling

We have a fairly complex set of loops around walking our list of
delalloc inodes when we find metadata delalloc space running low.
It doesn't work very well, can use large amounts of CPU and doesn't
do very efficient writeback.

This switches us to kick the bdi flusher threads instead.  All dirty
data in btrfs is accounted as delalloc data, so this is very similar
in terms of what it writes, but we're able to just kick off the IO
and wait for progress.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: tune the chunk allocation to 5% of the FS as metadata
Chris Mason [Tue, 26 Oct 2010 17:37:56 +0000 (13:37 -0400)]
Btrfs: tune the chunk allocation to 5% of the FS as metadata

An earlier commit tried to keep us from allocating too many
empty metadata chunks.  It was somewhat too restrictive and could
lead to ENOSPC errors on empty filesystems.

This increases the limits to about 5% of the FS size, allowing more
metadata chunks to be preallocated.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoAdd new functions for triggering inode writeback
Chris Mason [Fri, 29 Oct 2010 15:16:17 +0000 (11:16 -0400)]
Add new functions for triggering inode writeback

When btrfs is running low on metadata space, it needs to force delayed
allocation pages to disk.  It currently does this with a suboptimal walk
of a private list of inodes with delayed allocation, and it would be
much better if we used the generic flusher threads.

writeback_inodes_sb_if_idle would be ideal, but it waits for the flusher
thread to start IO on all the dirty pages in the FS before it returns.
This adds variants of writeback_inodes_sb* that allow the caller to
control how many pages get sent down.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: don't loop forever on bad btree blocks
Chris Mason [Sun, 24 Oct 2010 15:01:27 +0000 (11:01 -0400)]
Btrfs: don't loop forever on bad btree blocks

When btrfs discovers the generation number in a btree block is
incorrect, it can loop forever without forcing the RAID
code to try a valid mirror, and without returning EIO.

This changes things to properly kick out the EIO.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoMerge branch 'bug-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/josef/btrfs...
Chris Mason [Fri, 29 Oct 2010 13:27:49 +0000 (09:27 -0400)]
Merge branch 'bug-fixes' of git://git./linux/kernel/git/josef/btrfs-work

Conflicts:
fs/btrfs/extent-tree.c

Signed-off-by: Chris Mason <chris.mason@oracle.com>
14 years agoBtrfs: let the user know space caching is enabled
Josef Bacik [Thu, 28 Oct 2010 20:55:47 +0000 (16:55 -0400)]
Btrfs: let the user know space caching is enabled

If you mount -o space_cache, the option will be persistent across mounts, but to
make sure the user knows that they did this, emit a message telling them if they
didn't mount with -o space_cache but the feature is still used.

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: Add a clear_cache mount option
Josef Bacik [Tue, 21 Sep 2010 18:21:34 +0000 (14:21 -0400)]
Btrfs: Add a clear_cache mount option

If something goes wrong with the free space cache we need a way to make sure
it's not loaded on mount and that it's cleared for everybody.  When you pass the
clear_cache option it will make it so all block groups are setup to be cleared,
which keeps them from being loaded and then they will be truncated when the
transaction is committed.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: add support for mixed data+metadata block groups
Josef Bacik [Thu, 16 Sep 2010 20:19:09 +0000 (16:19 -0400)]
Btrfs: add support for mixed data+metadata block groups

There are just a few things that need to be fixed in the kernel to support mixed
data+metadata block groups.  Mostly we just need to make sure that if we are
using mixed block groups that we continue to allocate mixed block groups as we
need them.  Also we need to make sure __find_space_info will find our space info
if we search for DATA or METADATA only.  Tested this with xfstests and it works
nicely.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: check cache->caching_ctl before returning if caching has started
Josef Bacik [Thu, 16 Sep 2010 20:17:03 +0000 (16:17 -0400)]
Btrfs: check cache->caching_ctl before returning if caching has started

With the free space disk caching we can mark the block group as started with the
caching, but we don't have a caching ctl.  This can race with anybody else who
tries to get the caching ctl before we cache (this is very hard to do btw).  So
instead check to see if cache->caching_ctl is set, and if not return NULL.
Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: load free space cache if it exists
Josef Bacik [Wed, 25 Aug 2010 20:54:15 +0000 (16:54 -0400)]
Btrfs: load free space cache if it exists

This patch actually loads the free space cache if it exists.  The only thing
that really changes here is that we need to cache the block group if we're going
to remove an extent from it.  Previously we did not do this since the caching
kthread would pick it up.  With the on disk cache we don't have this luxury so
we need to make sure we read the on disk cache in first, and then remove the
extent, that way when the extent is unpinned the free space is added to the
block group.  This has been tested with all sorts of things.

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: write out free space cache
Josef Bacik [Fri, 2 Jul 2010 16:14:14 +0000 (12:14 -0400)]
Btrfs: write out free space cache

This is a simple bit, just dump the free space cache out to our preallocated
inode when we're writing out dirty block groups.  There are a bunch of changes
in inode.c in order to account for special cases.  Mostly when we're doing the
writeout we're holding trans_mutex, so we need to use the nolock transacation
functions.  Also we can't do asynchronous completions since the async thread
could be blocked on already completed IO waiting for the transaction lock.  This
has been tested with xfstests and btrfs filesystem balance, as well as my ENOSPC
tests.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: create special free space cache inode
Josef Bacik [Mon, 21 Jun 2010 18:48:16 +0000 (14:48 -0400)]
Btrfs: create special free space cache inode

In order to save free space cache, we need an inode to hold the data, and we
need a special item to point at the right inode for the right block group.  So
first, create a special item that will point to the right inode, and the number
of extent entries we will have and the number of bitmaps we will have.  We
truncate and pre-allocate space everytime to make sure it's uptodate.

This feature will be turned on as soon as you mount with -o space_cache, however
it is safe to boot into old kernels, they will just generate the cache the old
fashion way.  When you boot back into a newer kernel we will notice that we
modified and not the cache and automatically discard the cache.

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: remove warn_on from use_block_rsv
Josef Bacik [Tue, 26 Oct 2010 16:55:03 +0000 (12:55 -0400)]
Btrfs: remove warn_on from use_block_rsv

Because btrfs_dirty_inode does a btrfs_join_transaction, it doesn't actually
reserve space.  It does this so we can try and dirty the inode quickly without
having to deal with the ENOSPC problems.  But if it does get back ENOSPC it
handles it properly.  The problem is use_block_rsv does a WARN_ON whenever this
case happens, even tho btrfs_dirty_inode takes it into account and actually
expects to get -ENOSPC if things are particularly tight.  So instead just remove
the warning.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: set trans to null in reserve_metadata_bytes if we commit the transaction
Josef Bacik [Tue, 26 Oct 2010 16:52:53 +0000 (12:52 -0400)]
Btrfs: set trans to null in reserve_metadata_bytes if we commit the transaction

btrfs_commit_transaction will free our trans, but because we pass trans to
shrink_delalloc we could possibly have a use after free situation.  So instead
if we commit the transaction, set trans to null and set committed to true so we
don't keep trying to commit a transaction.  This fixes a panic I could reproduce
at will.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: fix error handling in btrfs_get_sb
Josef Bacik [Fri, 22 Oct 2010 19:26:53 +0000 (15:26 -0400)]
Btrfs: fix error handling in btrfs_get_sb

If we failed to find the root subvol id, or the subvol=<name>, we would
deactivate the locked super and close the devices.  The problem is at this point
we have gotten the SB all setup, which includes setting super_operations, so
when we'd deactiveate the super, we'd do a close_ctree() which closes the
devices, so we'd end up closing the devices twice.  So if you do something like
this

mount /dev/sda1 /mnt/test1
mount /dev/sda1 /mnt/test2 -o subvol=xxx
umount /mnt/test1

it would blow up (if subvol xxx doesn't exist).  This patch fixes that problem.
Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: rework how we reserve metadata bytes
Josef Bacik [Fri, 15 Oct 2010 20:52:49 +0000 (16:52 -0400)]
Btrfs: rework how we reserve metadata bytes

With multi-threaded writes we were getting ENOSPC early because somebody would
come in, start flushing delalloc because they couldn't make their reservation,
and in the meantime other threads would come in and use the space that was
getting freed up, so when the original thread went to check to see if they had
space they didn't and they'd return ENOSPC.  So instead if we have some free
space but not enough for our reservation, take the reservation and then start
doing the flushing.  The only time we don't take reservations is when we've
already overcommitted our space, that way we don't have people who come late to
the party way overcommitting ourselves.  This also moves all of the retrying and
flushing code into reserve_metdata_bytes so it's all uniform.  This keeps my
fs_mark test from returning -ENOSPC as soon as it starts and actually lets me
fill up the disk.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: don't allocate chunks as aggressively
Josef Bacik [Fri, 15 Oct 2010 19:23:48 +0000 (15:23 -0400)]
Btrfs: don't allocate chunks as aggressively

Because the ENOSPC code over reserves super aggressively we end up allocating
chunks way more often than we should.  For example with my fs_mark tests on a
2gb fs I can end up reserved 1gb just for metadata, when only 34mb of that is
being used.  So instead check to see if the amount of space actually used is
less than 30% of the total space, and if so don't allocate a chunk, but only if
we have at least 256mb of free space to make sure we don't put too much pressure
on free space.

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: re-work delalloc flushing
Josef Bacik [Fri, 15 Oct 2010 19:18:40 +0000 (15:18 -0400)]
Btrfs: re-work delalloc flushing

Currently we try and flush delalloc, but we only do that in a sort of weak way,
which works fine in most cases but if we're under heavy pressure we need to be
able to wait for flushing to happen.  Also instead of checking the bytes
reserved in the block_rsv, check the space info since it is more accurate.  The
sync option will be used in a future patch.

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: fix reservation code for mixed block groups
Josef Bacik [Fri, 15 Oct 2010 19:13:32 +0000 (15:13 -0400)]
Btrfs: fix reservation code for mixed block groups

The global reservation stuff tries to add together DATA and METADATA used in
order to figure out how much to reserve for everything, but this doesn't work
right for mixed block groups.  Instead if we have mixed block groups just set
data used to 0.  Also with mixed block groups we will use bytes_may_use for
keeping track of delalloc bytes, so we need to take that into account in our
reservation calculations.

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: fix df regression
Josef Bacik [Thu, 14 Oct 2010 18:52:27 +0000 (14:52 -0400)]
Btrfs: fix df regression

The new ENOSPC stuff breaks out the raid types which breaks the way we were
reporting df to the system.  This fixes it back so that Available is the total
space available to data and used is the actual bytes used by the filesystem.
This means that Available is Total - data used - all of the metadata space.
Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: fix the df ioctl to report raid types
Josef Bacik [Wed, 29 Sep 2010 15:22:36 +0000 (11:22 -0400)]
Btrfs: fix the df ioctl to report raid types

The new ENOSPC stuff broke the df ioctl since we no longer create seperate space
info's for each RAID type.  So instead, loop through each space info's raid
lists so we can get the right RAID information which will allow the df ioctl to
tell us RAID types again.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoBtrfs: stop trying to shrink delalloc if there are no inodes to reclaim
Josef Bacik [Thu, 16 Sep 2010 18:29:55 +0000 (14:29 -0400)]
Btrfs: stop trying to shrink delalloc if there are no inodes to reclaim

In very severe ENOSPC cases we can run out of inodes to do delalloc on, which
means we'll just keep looping trying to shrink delalloc.  Instead, if we fail to
shrink delalloc 3 times in a row break out since we're not likely to make any
progress.  Tested this with a 100mb fs an xfstests test 13.  Before the patch it
would hang the box, with the patch we get -ENOSPC like we should.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
14 years agoLinux 2.6.36
Linus Torvalds [Wed, 20 Oct 2010 20:30:22 +0000 (13:30 -0700)]
Linux 2.6.36

14 years agoMerge branch 'upstream' of git://git.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Wed, 20 Oct 2010 20:18:21 +0000 (13:18 -0700)]
Merge branch 'upstream' of git://git.linux-mips.org/upstream-linus

* 'upstream' of git://git.linux-mips.org/pub/scm/upstream-linus:
  MIPS: O32 compat/N32: Fix to use compat syscall wrappers for AIO syscalls.
  MAINTAINERS: Change list for ioc_serial to linux-serial.
  SERIAL: ioc3_serial: Return -ENOMEM on memory allocation failure
  MIPS: jz4740: Fix Kbuild Platform file.
  MIPS: Repair Kbuild make clean breakage.

14 years agovirtio: console: Don't block entire guest if host doesn't read data
Amit Shah [Wed, 20 Oct 2010 03:15:43 +0000 (13:45 +1030)]
virtio: console: Don't block entire guest if host doesn't read data

If the host is slow in reading data or doesn't read data at all,
blocking write calls not only blocked the program that called write()
but the entire guest itself.

To overcome this, let's not block till the host signals it has given
back the virtio ring element we passed it.  Instead, send the buffer to
the host and return to userspace.  This operation then becomes similar
to how non-blocking writes work, so let's use the existing code for this
path as well.

This code change also ensures blocking write calls do get blocked if
there's not enough room in the virtio ring as well as they don't return
-EAGAIN to userspace.

Signed-off-by: Amit Shah <amit.shah@redhat.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
CC: stable@kernel.org
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Wed, 20 Oct 2010 20:13:09 +0000 (13:13 -0700)]
Merge git://git./linux/kernel/git/jejb/scsi-rc-fixes-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] bsg: fix incorrect device_status value
  [SCSI] Fix VPD inquiry page wrapper

14 years agoMerge branch 'kvm-updates/2.6.36' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Wed, 20 Oct 2010 16:00:44 +0000 (09:00 -0700)]
Merge branch 'kvm-updates/2.6.36' of git://git./virt/kvm/kvm

* 'kvm-updates/2.6.36' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: Fix fs/gs reload oops with invalid ldt

14 years agoMIPS: O32 compat/N32: Fix to use compat syscall wrappers for AIO syscalls.
Michel Thebeau [Tue, 28 Sep 2010 18:15:37 +0000 (14:15 -0400)]
MIPS: O32 compat/N32: Fix to use compat syscall wrappers for AIO syscalls.

[Ralf: Michel's original patch only fixed N32; I replicated the same fix
for O32.]

Signed-off-by: Michel Thebeau <michel.thebeau@windriver.com>
Cc: paul.gortmaker@windriver.com
Cc: bruce.ashfield@windriver.com
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMAINTAINERS: Change list for ioc_serial to linux-serial.
Ralf Baechle [Tue, 19 Oct 2010 17:32:41 +0000 (18:32 +0100)]
MAINTAINERS: Change list for ioc_serial to linux-serial.

IOC3 is also being used on SGI MIPS systems but this particular driver is
only being used on IA64 systems so linux-mips made no sense as a list.  Pat
also thinks linux-serial@vger.kernel.org is the better list.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoSERIAL: ioc3_serial: Return -ENOMEM on memory allocation failure
Julia Lawall [Fri, 15 Oct 2010 13:00:09 +0000 (15:00 +0200)]
SERIAL: ioc3_serial: Return -ENOMEM on memory allocation failure

In this code, 0 is returned on memory allocation failure, even though other
failures return -ENOMEM or other similar values.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
expression ret;
expression x,e1,e2,e3;
@@

ret = 0
... when != ret = e1
*x = \(kmalloc\|kcalloc\|kzalloc\)(...)
... when != ret = e2
if (x == NULL) { ... when != ret = e3
  return ret;
}
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
To: Pat Gefre <pfg@sgi.com>
Cc: kernel-janitors@vger.kernel.org
Cc: linux-ia64@vger.kernel.org
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/1704/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: jz4740: Fix Kbuild Platform file.
David Daney [Tue, 19 Oct 2010 01:05:49 +0000 (18:05 -0700)]
MIPS: jz4740: Fix Kbuild Platform file.

The platform specific files should be included via the platform-y
variable.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Cc: Lars-Peter Clausen <lars@metafoo.de>
Patchwork: https://patchwork.linux-mips.org/patch/1719/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Repair Kbuild make clean breakage.
David Daney [Tue, 19 Oct 2010 00:51:26 +0000 (17:51 -0700)]
MIPS: Repair Kbuild make clean breakage.

When running make clean, Kbuild doesn't process the .config file, so nothing
generates a platform-y variable.  We can get it to descend into the platform
directories by setting $(obj-).

The dec Platform file was unconditionally setting platform-, obliterating
its previous contents and preventing some directories from being cleaned.
This is change to an append operation '+=' to allow cavium-octeon to be
cleaned.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Patchwork: https://patchwork.linux-mips.org/patch/1718/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMerge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Tue, 19 Oct 2010 17:10:20 +0000 (10:10 -0700)]
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/radeon/kms: avivo cursor workaround applies to evergreen as well

14 years agoKVM: Fix fs/gs reload oops with invalid ldt
Avi Kivity [Tue, 19 Oct 2010 14:46:55 +0000 (16:46 +0200)]
KVM: Fix fs/gs reload oops with invalid ldt

kvm reloads the host's fs and gs blindly, however the underlying segment
descriptors may be invalid due to the user modifying the ldt after loading
them.

Fix by using the safe accessors (loadsegment() and load_gs_index()) instead
of home grown unsafe versions.

This is CVE-2010-3698.

KVM-Stable-Tag.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
14 years agoMerge branch 'upstream' of git://git.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Mon, 18 Oct 2010 20:10:36 +0000 (13:10 -0700)]
Merge branch 'upstream' of git://git.linux-mips.org/upstream-linus

* 'upstream' of git://git.linux-mips.org/pub/scm/upstream-linus:
  MIPS: Enable ISA_DMA_API config to fix build failure
  MIPS: 32-bit: Fix build failure in asm/fcntl.h
  MIPS: Remove all generated vmlinuz* files on "make clean"
  MIPS: do_sigaltstack() expects userland pointers
  MIPS: Fix error values in case of bad_stack
  MIPS: Sanitize restart logics
  MIPS: secure_computing, syscall audit: syscall number should in r2, not r0.
  MIPS: Don't block signals if we'd failed to setup a sigframe

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Mon, 18 Oct 2010 20:10:08 +0000 (13:10 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: evdev - fix EVIOCSABS regression
  Input: evdev - fix Ooops in EVIOCGABS/EVIOCSABS

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394...
Linus Torvalds [Mon, 18 Oct 2010 20:09:26 +0000 (13:09 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ieee1394/linux1394-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6:
  firewire: ohci: fix TI TSB82AA2 regression since 2.6.35

14 years agomxc_nand: do not depend on disabling the irq in the interrupt handler
Sascha Hauer [Mon, 18 Oct 2010 08:16:26 +0000 (10:16 +0200)]
mxc_nand: do not depend on disabling the irq in the interrupt handler

This patch reverts the driver to enabling/disabling the NFC interrupt
mask rather than enabling/disabling the system interrupt.  This cleans
up the driver so that it doesn't rely on interrupts being disabled
within the interrupt handler.

For i.MX21 we keep the current behaviour, that is calling
enable_irq/disable_irq_nosync to enable/disable interrupts.  This patch
is based on earlier work by John Ogness.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Acked-by: John Ogness <john.ogness@linutronix.de>
Tested-by: John Ogness <john.ogness@linutronix.de>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge branch 'for-linus/i2c/2636-rc8' of git://git.fluff.org/bjdooks/linux
Linus Torvalds [Mon, 18 Oct 2010 20:05:10 +0000 (13:05 -0700)]
Merge branch 'for-linus/i2c/2636-rc8' of git://git.fluff.org/bjdooks/linux

* 'for-linus/i2c/2636-rc8' of git://git.fluff.org/bjdooks/linux:
  i2c-imx: do not allow interruptions when waiting for I2C to complete
  i2c-davinci: Fix TX setup for more SoCs

14 years agoMerge branch 'fixes'
Linus Torvalds [Mon, 18 Oct 2010 20:04:33 +0000 (13:04 -0700)]
Merge branch 'fixes'

* fixes:
  v4l1: fix 32-bit compat microcode loading translation
  De-pessimize rds_page_copy_user

14 years agoMIPS: Enable ISA_DMA_API config to fix build failure
Namhyung Kim [Mon, 18 Oct 2010 03:55:21 +0000 (12:55 +0900)]
MIPS: Enable ISA_DMA_API config to fix build failure

Add ISA_DMA_API config item and select it when GENERIC_ISA_DMA enabled.
This fixes build failure on allmodconfig like following:

  CC      sound/isa/es18xx.o
sound/isa/es18xx.c: In function 'snd_es18xx_playback1_prepare':
sound/isa/es18xx.c:501:9: error: implicit declaration of function 'snd_dma_program'
sound/isa/es18xx.c: In function 'snd_es18xx_playback_pointer':
sound/isa/es18xx.c:818:3: error: implicit declaration of function 'snd_dma_pointer'
make[3]: *** [sound/isa/es18xx.o] Error 1
make[2]: *** [sound/isa/es18xx.o] Error 2
make[1]: *** [sub-make] Error 2
make: *** [all] Error 2

Signed-off-by: Namhyung Kim <namhyung@gmail.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/1717/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: 32-bit: Fix build failure in asm/fcntl.h
Namhyung Kim [Sun, 17 Oct 2010 16:41:39 +0000 (01:41 +0900)]
MIPS: 32-bit: Fix build failure in asm/fcntl.h

  CC      security/integrity/ima/ima_fs.o
In file included from linux/include/linux/fcntl.h:4:0,
                 from linux/security/integrity/ima/ima_fs.c:18:
linux/arch/mips/include/asm/fcntl.h:63:2: error: expected specifier-qualifier-list before 'off_t'
make[3]: *** [security/integrity/ima/ima_fs.o] Error 1
make[2]: *** [security/integrity/ima/ima_fs.o] Error 2
make[1]: *** [sub-make] Error 2
make: *** [all] Error 2

Signed-off-by: Namhyung Kim <namhyung@gmail.com>
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/1715/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Remove all generated vmlinuz* files on "make clean"
Wu Zhangjin [Wed, 22 Sep 2010 06:03:57 +0000 (14:03 +0800)]
MIPS: Remove all generated vmlinuz* files on "make clean"

[Ralf: I changed the patch to explicitly list all files to be deleted out
of paranoia.]

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1590/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: do_sigaltstack() expects userland pointers
Al Viro [Tue, 28 Sep 2010 17:50:57 +0000 (18:50 +0100)]
MIPS: do_sigaltstack() expects userland pointers

o32 compat does the right thing, native and n32 compat do not...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arch@vger.kernel.org
Cc: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1700/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Fix error values in case of bad_stack
Al Viro [Tue, 28 Sep 2010 17:50:47 +0000 (18:50 +0100)]
MIPS: Fix error values in case of bad_stack

We want EFAULT, not -<syscall number>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arch@vger.kernel.org
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/1699/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Sanitize restart logics
Al Viro [Tue, 28 Sep 2010 17:50:37 +0000 (18:50 +0100)]
MIPS: Sanitize restart logics

Put the original syscall number into ->regs[0] when we leave syscall
with error.  Use it in restart logics.  Everything else will have
it 0 since we pass through SAVE_SOME on all the ways in.  Note that
in places like bad_stack and inllegal_syscall we leave it 0 - it's not
restartable.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arch@vger.kernel.org
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/1698/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: secure_computing, syscall audit: syscall number should in r2, not r0.
Al Viro [Tue, 28 Sep 2010 17:50:27 +0000 (18:50 +0100)]
MIPS: secure_computing, syscall audit: syscall number should in r2, not r0.

As it is, audit_syscall_entry() and secure_computing() get the
bogus value (0, in fact)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arch@vger.kernel.org
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/1697/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Don't block signals if we'd failed to setup a sigframe
Al Viro [Tue, 28 Sep 2010 17:50:17 +0000 (18:50 +0100)]
MIPS: Don't block signals if we'd failed to setup a sigframe

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arch@vger.kernel.org
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/1696/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoInput: evdev - fix EVIOCSABS regression
Daniel Mack [Mon, 18 Oct 2010 15:43:50 +0000 (08:43 -0700)]
Input: evdev - fix EVIOCSABS regression

448cd16 ("Input: evdev - rearrange ioctl handling") broke EVIOCSABS by
checking for the wrong direction bit.

Signed-off-by: Daniel Mack <zonque@gmail.com>
Reported-by: Sven Neumann <s.neumann@raumfeld.com>
Tested-by: Sven Neumann <s.neumann@raumfeld.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
14 years agoInput: evdev - fix Ooops in EVIOCGABS/EVIOCSABS
Daniel Mack [Mon, 18 Oct 2010 15:43:30 +0000 (08:43 -0700)]
Input: evdev - fix Ooops in EVIOCGABS/EVIOCSABS

This fixes a regression introduced by the dynamic allocation of absinfo
for input devices. We need to bail out early for input devices which
don't have absolute axis.

[  929.664303] Pid: 2989, comm: input Not tainted 2.6.36-rc8+ #14 MS-7260/MS-7260
[  929.664318] EIP: 0060:[<c12bdc01>] EFLAGS: 00010246 CPU: 0
[  929.664331] EIP is at evdev_ioctl+0x4f8/0x59f
[  929.664341] EAX: 00000040 EBX: 00000000 ECX: 00000006 EDX: f45a1efc
[  929.664355] ESI: 00000000 EDI: f45a1efc EBP: f45a1f24 ESP: f45a1eb8
[  929.664369]  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
[  929.664402]  f470da74 f6a30e78 f462c240 00000018 bfe4a260 00000000 f45b06fc 00000000
[  929.664429] <0> 000000c4 b769d000 c3544620 f470da74 f45b06fc f45b06fc f45a1f38 c107dd1f
[  929.664458] <0> f4710b74 000000c4 00000000 00000000 00000000 0000029d 00000a74 f4710b74
[  929.664500]  [<c107dd1f>] ? handle_mm_fault+0x2be/0x59a
[  929.664513]  [<c12bd709>] ? evdev_ioctl+0x0/0x59f
[  929.664524]  [<c1099d30>] ? do_vfs_ioctl+0x494/0x4d9
[  929.664538]  [<c10432a1>] ? up_read+0x16/0x29
[  929.664550]  [<c101c818>] ? do_page_fault+0x2ff/0x32d
[  929.664564]  [<c108d048>] ? do_sys_open+0xc5/0xcf
[  929.664575]  [<c1099db6>] ? sys_ioctl+0x41/0x61
[  929.664587]  [<c1002710>] ? sysenter_do_call+0x12/0x36
[  929.684570] ---[ end trace 11b83e923bd8f2bb ]---

Signed-off-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
14 years agoi2c-imx: do not allow interruptions when waiting for I2C to complete
Marc Kleine-Budde [Mon, 21 Jun 2010 07:27:05 +0000 (09:27 +0200)]
i2c-imx: do not allow interruptions when waiting for I2C to complete

The i2c_imx_trx_complete() function is using
wait_event_interruptible_timeout() to wait for the I2C controller to
signal that it has completed an I2C bus operation. If the process that
causes the I2C operation receives a signal, the wait will be
interrupted, returning an error. It is better to let the I2C operation
finished before handling the signal (i.e. returning into userspace).

It is safe to use wait_event_timeout() instead, because the timeout
will allow the process to exit if the I2C bus hangs. It's also better
to allow the I2C operation to finish, because unacknowledged I2C
operations can cause the I2C bus to hang.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Reviewed-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
14 years agoi2c-davinci: Fix TX setup for more SoCs
Jon Povey [Tue, 12 Oct 2010 04:47:05 +0000 (13:47 +0900)]
i2c-davinci: Fix TX setup for more SoCs

This patch is an improvement to 4bba0fd8d1c6d405df666e2573e1a1f917098be0
which got to mainline a little early.

Sudhakar Rajashekhara explains that at least OMAP-L138 requires MDR mode
settings before DXR for correct behaviour, so load MDR first with
STT cleared and later load again with STT set.

Tested on DM355 connected to Techwell TW2836 and Wolfson WM8985

Signed-off-by: Jon Povey <jon.povey@racelogic.co.uk>
Acked-by: Troy Kisky <troy.kisky@boundarydevices.com>
Tested-by: Sudhakar Rajashekhara <sudhakar.raj@ti.com>
Acked-by: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Ben Dooks <ben-linux@fluff.org>