unionfs-2.6.23.y.git
17 years agoCommenting the code.
Erez Zadok [Sun, 25 Mar 2007 04:27:12 +0000 (00:27 -0400)]
Commenting the code.

Audited entire code for documentation.  Added comments at top of functions
where it felt necessary (i.e., function's name and size don't make it clear
what it may be doing precisely).  Reformatted some long comments.  Fixed a
few comment typos and spelling errors.

17 years agocleanup: prefix external functions with 'extern' properly
Erez Zadok [Sun, 25 Mar 2007 03:14:03 +0000 (23:14 -0400)]
cleanup: prefix external functions with 'extern' properly

17 years agocleanup: one more trailing whitespace removed.
Erez_Zadok [Sun, 25 Mar 2007 03:01:13 +0000 (23:01 -0400)]
cleanup: one more trailing whitespace removed.

17 years agocleanup: spell check all strings and comments
Erez_Zadok [Sun, 25 Mar 2007 00:06:44 +0000 (20:06 -0400)]
cleanup: spell check all strings and comments

17 years agocleanup: use consistent style for multi-line comments
Erez_Zadok [Sat, 24 Mar 2007 23:56:56 +0000 (19:56 -0400)]
cleanup: use consistent style for multi-line comments

17 years agocleanup: keep all top-of-function automatic variables clustered together
Erez_Zadok [Sat, 24 Mar 2007 23:41:58 +0000 (19:41 -0400)]
cleanup: keep all top-of-function automatic variables clustered together

17 years agocoding-style: remove one more {braces} around one-statement code block
Erez_Zadok [Sat, 24 Mar 2007 23:32:42 +0000 (19:32 -0400)]
coding-style: remove one more {braces} around one-statement code block

17 years agocleanup: remove unnecessary blank lines in short code blocks
Erez_Zadok [Sat, 24 Mar 2007 23:31:28 +0000 (19:31 -0400)]
cleanup: remove unnecessary blank lines in short code blocks

17 years agocleanup: add missing copyright notices
Erez_Zadok [Sat, 24 Mar 2007 23:21:51 +0000 (19:21 -0400)]
cleanup: add missing copyright notices

17 years agocoding-style: reformat lines longer than 80 chars (accounting for TABs)
Erez_Zadok [Sat, 24 Mar 2007 23:14:18 +0000 (19:14 -0400)]
coding-style: reformat lines longer than 80 chars (accounting for TABs)

17 years agocleanup: every printk should prefix with "unionfs: " consistently
Erez_Zadok [Sat, 24 Mar 2007 22:36:59 +0000 (18:36 -0400)]
cleanup: every printk should prefix with "unionfs: " consistently

17 years agocoding-style: remove unnecessary {braces} around one-line code blocks
Erez_Zadok [Sat, 24 Mar 2007 22:19:49 +0000 (18:19 -0400)]
coding-style: remove unnecessary {braces} around one-line code blocks

17 years agocoding-style: no lines over 80 chars
Erez_Zadok [Sat, 24 Mar 2007 22:11:10 +0000 (18:11 -0400)]
coding-style: no lines over 80 chars

17 years agocoding-style: indent every line properly
Erez_Zadok [Sat, 24 Mar 2007 21:57:56 +0000 (17:57 -0400)]
coding-style: indent every line properly

17 years agocleanup: remove trailing spaces
Erez_Zadok [Sat, 24 Mar 2007 21:41:57 +0000 (17:41 -0400)]
cleanup: remove trailing spaces

17 years agoDocumentation: fix spelling error
Erez_Zadok [Sat, 24 Mar 2007 20:28:59 +0000 (16:28 -0400)]
Documentation: fix spelling error

17 years agoDocumentation: minor typo fixed in incgen remount option
Erez_Zadok [Sat, 24 Mar 2007 20:19:40 +0000 (16:19 -0400)]
Documentation: minor typo fixed in incgen remount option

17 years agoUnionfs: SLAB_CTOR_CONSTRUCTOR has been removed from vanilla
Josef 'Jeff' Sipek [Sat, 19 May 2007 07:14:30 +0000 (03:14 -0400)]
Unionfs: SLAB_CTOR_CONSTRUCTOR has been removed from vanilla

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years ago[PATCH] unionfs: fix slab abuses with krealloc
Pekka Enberg [Sun, 13 May 2007 16:55:44 +0000 (12:55 -0400)]
[PATCH] unionfs: fix slab abuses with krealloc

This changes unionfs to use krealloc() for reallocating memory so that we
don't need to play tricks with slab internals.

Cc: Josef Sipek <jsipek@fsl.cs.sunysb.edu>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years ago[PATCH] fix unionfs compilation
Adrian Bunk [Sat, 5 May 2007 19:28:33 +0000 (15:28 -0400)]
[PATCH] fix unionfs compilation

On Sat, May 05, 2007 at 01:49:55AM -0700, Andrew Morton wrote:
>...
> Changes since 2.6.21-rc7-mm2:
>...
>  git-unionfs.patch
>...
>  git trees
>...

<--  snip  -->

...
  CC      fs/unionfs/super.o
/home/bunk/linux/kernel-2.6/linux-2.6.21-mm1/fs/unionfs/super.c: In function ‘init_once’:
/home/bunk/linux/kernel-2.6/linux-2.6.21-mm1/fs/unionfs/super.c:822: error: ‘SLAB_CTOR_VERIFY’ undeclared (first use in this function)
/home/bunk/linux/kernel-2.6/linux-2.6.21-mm1/fs/unionfs/super.c:822: error: (Each undeclared identifier is reported only once
/home/bunk/linux/kernel-2.6/linux-2.6.21-mm1/fs/unionfs/super.c:822: error: for each function it appears in.)
make[3]: *** [fs/unionfs/super.o] Error 1

<--  snip  -->

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Check remount options for being NULL
Adrian Brunyate [Thu, 26 Apr 2007 22:51:16 +0000 (18:51 -0400)]
Unionfs: Check remount options for being NULL

Signed-off-by: Adrian Brunyate <abrunyate@yahoo.com>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Accept MS_SILENT during remount
Adrian Brunyate [Thu, 26 Apr 2007 22:51:09 +0000 (18:51 -0400)]
Unionfs: Accept MS_SILENT during remount

[jsipek: whitespace cleanup]
Signed-off-by: Adrian Brunyate <abrunyate@yahoo.com>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Don't inline do_remount_{add,del,mode}_option
Erez Zadok [Fri, 23 Mar 2007 17:10:46 +0000 (13:10 -0400)]
Unionfs: Don't inline do_remount_{add,del,mode}_option

gcc4 decided to inline do_remount_{add,del,mode}_option creating an 600 byte
stack abuser on a x86_64 test box.

Reported by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Added several BUG_ONs to assert dentry validity
Erez Zadok [Fri, 23 Mar 2007 17:07:14 +0000 (13:07 -0400)]
Unionfs: Added several BUG_ONs to assert dentry validity

This should help catch races between the VFS and the branch-management code.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Properly handle stale inodes passed to unionfs_permission
Erez Zadok [Fri, 23 Mar 2007 00:17:48 +0000 (20:17 -0400)]
Unionfs: Properly handle stale inodes passed to unionfs_permission

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Pass lowernd to lower ->revalidate function
Erez Zadok [Fri, 23 Mar 2007 00:09:07 +0000 (20:09 -0400)]
Unionfs: Pass lowernd to lower ->revalidate function

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: vfsmount reference counting fixes
Erez Zadok [Fri, 23 Mar 2007 00:08:30 +0000 (20:08 -0400)]
Unionfs: vfsmount reference counting fixes

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: unionfs_create needs to revalidate the dentry
Erez Zadok [Fri, 23 Mar 2007 00:01:56 +0000 (20:01 -0400)]
Unionfs: unionfs_create needs to revalidate the dentry

We have to read-lock the superblock rwsem, and we have to revalidate the
parent dentry and this one.  A branch-management operation could have taken
place, mid-way through a VFS operation that eventually reaches
unionfs_create().  So we have to ensure consistency, just as we do with the
file operations.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Decrement totalopens counter on error in unionfs_open
Erez Zadok [Thu, 22 Mar 2007 23:47:43 +0000 (19:47 -0400)]
Unionfs: Decrement totalopens counter on error in unionfs_open

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Document unionfs_d_release locking
Josef 'Jeff' Sipek [Sun, 18 Mar 2007 20:30:45 +0000 (16:30 -0400)]
Unionfs: Document unionfs_d_release locking

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Remove the older incgen ioctl
Josef 'Jeff' Sipek [Sun, 18 Mar 2007 20:22:03 +0000 (16:22 -0400)]
Unionfs: Remove the older incgen ioctl

The new remount code now has the "incgen" functionality.

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Grab the unionfs sb private data lock around branch info users
Erez Zadok [Thu, 22 Mar 2007 23:32:40 +0000 (19:32 -0400)]
Unionfs: Grab the unionfs sb private data lock around branch info users

Locking/concurrency/race fixes.  Use the unionfs superblock rwsem, and grab
the read lock around every op that uses branch-related information, such as
branch counters.  Grab the write rwsem lock in operations which attempt to
change branch information, such as when adding/deleting branches.  This
will, for example, cause branch-management remount commands (which are
infrequent) to block a bit until all in-progress file operations on open
files are done.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
[jsipek: whitespace fixes & more locks/unlocks]
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Rewrite unionfs_d_revalidate
Erez Zadok [Tue, 20 Mar 2007 01:02:55 +0000 (21:02 -0400)]
Unionfs: Rewrite unionfs_d_revalidate

Rewrite unionfs_d_revalidate code to avoid stack-unfriendly recursion: split
into a call to revalidate just one dentry, and an interative driver function
to revalidate an entire dentry-parent chain.

Fix vfsmount ref leaks which prevented lower f/s from being unmounted after
generation increment, esp. during heavy loads.

Fix one deadlock between revalidation code and VFS.

Better documentation of what the code does.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
[jsipek: compile & whitespace fixes]
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Introduce unionfs_mnt{get,put}
Erez Zadok [Tue, 20 Mar 2007 00:57:49 +0000 (20:57 -0400)]
Unionfs: Introduce unionfs_mnt{get,put}

Helper inline functions to perform Unionfs's mntget/put ops on lower
branches.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
[jsipek: cleanup branching in unionfs_mnt{get,put} and compile fixes]
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Bulk of branch-management remount code
Erez Zadok [Tue, 20 Mar 2007 00:44:12 +0000 (20:44 -0400)]
Unionfs: Bulk of branch-management remount code

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Introduce branch-id code
Erez Zadok [Thu, 22 Mar 2007 23:43:29 +0000 (19:43 -0400)]
Unionfs: Introduce branch-id code

Each branch gets a unique ID, which helps during branch additions,
deletions, and changes, to locate where branches were moved to, and perform
proper reference-counting.  This is useful even if the same directory was
added more than once to union.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Actually verify if dentry's info node is locked
Erez Zadok [Mon, 9 Apr 2007 13:57:51 +0000 (09:57 -0400)]
Unionfs: Actually verify if dentry's info node is locked

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Provide more helpful info on branch leaks during unmount
Erez Zadok [Mon, 19 Mar 2007 07:01:49 +0000 (03:01 -0400)]
Unionfs: Provide more helpful info on branch leaks during unmount

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
[jsipek: no need to take a read lock on the superblock private data]
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Rename unionfs_data sbcount field to more appropriate open_files
Erez Zadok [Mon, 19 Mar 2007 04:46:03 +0000 (00:46 -0400)]
Unionfs: Rename unionfs_data sbcount field to more appropriate open_files

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Proper comment on rwsem field
Josef 'Jeff' Sipek [Sun, 18 Mar 2007 23:37:51 +0000 (19:37 -0400)]
Unionfs: Proper comment on rwsem field

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Documentation updates for branch-management
Erez Zadok [Fri, 23 Mar 2007 21:17:07 +0000 (17:17 -0400)]
Unionfs: Documentation updates for branch-management

Describe dynamic branch-management introduced by subsequent patches.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs: Export drop_pagecache_sb symbol
Erez Zadok [Sun, 18 Mar 2007 23:24:29 +0000 (19:24 -0400)]
fs: Export drop_pagecache_sb symbol

Signed-off-by: Erez Zadok <ezkcs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipekcs.sunysb.edu>
17 years agofs: Introduce path{get,put}
Erez Zadok [Sun, 18 Mar 2007 23:23:52 +0000 (19:23 -0400)]
fs: Introduce path{get,put}

Export drop_pagecache_sb symbol (for branch-management).

Signed-off-by: Erez Zadok <ezkcs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipekcs.sunysb.edu>
17 years ago[PATCH] Unionfs: sioq not __exit
Randy Dunlap [Mon, 19 Mar 2007 23:15:22 +0000 (19:15 -0400)]
[PATCH] Unionfs: sioq not __exit

stop_sioq() is called from both __init and __exit functions, so it
shouldn't be marked __exit.

Reported on the kernelnewbies mailing list, but no patch offered there.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoMAINTAINERS: Unionfs URL update
Josef 'Jeff' Sipek [Sat, 17 Mar 2007 23:42:25 +0000 (19:42 -0400)]
MAINTAINERS: Unionfs URL update

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Fix unlocking in error paths
Erez Zadok [Sun, 4 Mar 2007 23:02:11 +0000 (18:02 -0500)]
fs/unionfs/: Fix unlocking in error paths

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Fix dentry leak in copyup_named_dentry
Erez Zadok [Sun, 4 Mar 2007 20:21:14 +0000 (15:21 -0500)]
fs/unionfs/: Fix dentry leak in copyup_named_dentry

When we chmod a directory on a readonly branch, and have to copy it up, we
forget to dput(). If this was a file, it gets dput indirectly through other
functions we call, but not if it was a directory.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: mntput in __cleanup_dentry
Erez Zadok [Fri, 2 Mar 2007 18:10:56 +0000 (13:10 -0500)]
fs/unionfs/: mntput in __cleanup_dentry

This fixes a mnt refleak which occured during copyup when directory
hierarchy was recreated on a writable branch.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Fix copyup_deleted_file dentry leak
Erez Zadok [Sun, 4 Mar 2007 05:41:51 +0000 (00:41 -0500)]
fs/unionfs/: Fix copyup_deleted_file dentry leak

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Miscellaneous coding style fixes
Josef 'Jeff' Sipek [Fri, 2 Mar 2007 18:14:01 +0000 (13:14 -0500)]
fs/unionfs/: Miscellaneous coding style fixes

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Check return value of d_path
Josef 'Jeff' Sipek [Tue, 27 Feb 2007 19:27:25 +0000 (14:27 -0500)]
fs/unionfs/: Check return value of d_path

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Use SEEK_{SET,CUR} instead of hardcoded values
Josef 'Jeff' Sipek [Sun, 25 Feb 2007 22:32:46 +0000 (17:32 -0500)]
fs/unionfs/: Use SEEK_{SET,CUR} instead of hardcoded values

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Remove alloc_filldir_node
Josef 'Jeff' Sipek [Sun, 25 Feb 2007 22:30:34 +0000 (17:30 -0500)]
fs/unionfs/: Remove alloc_filldir_node

alloc_filldir_node was used only once. Additionally, all the arguments
passed to it were ignored wasting stack space for no reason whatsoever.

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Rename unionfs_d_revalidate_wrap
Josef 'Jeff' Sipek [Sat, 24 Feb 2007 21:47:45 +0000 (16:47 -0500)]
fs/unionfs/: Rename unionfs_d_revalidate_wrap

Follow the convention of "foo" calling "__foo".

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Several small cleanups in unionfs_interpose
Josef 'Jeff' Sipek [Sat, 24 Feb 2007 21:46:08 +0000 (16:46 -0500)]
fs/unionfs/: Several small cleanups in unionfs_interpose

1) No need to lock the inode - lockdep was complaining about potential
circular dependency

2) No need to use temporary variable for iunique() inode number

3) Removed unneeded comment

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Don't grab dentry private data mutex in unionfs_d_release
Josef 'Jeff' Sipek [Fri, 23 Feb 2007 06:04:09 +0000 (01:04 -0500)]
fs/unionfs/: Don't grab dentry private data mutex in unionfs_d_release

Grabbing the UNIONFS_D(dentry)->lock is completely unnecessary and there are
no other references; we are about to free the object anyway. Additionally,
grabbing the mutex produces warning when the slab object is reused - as it
was freed while there still was a reference to it.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Fix a memory leak in unionfs_read_super
Erez Zadok [Fri, 2 Mar 2007 18:35:33 +0000 (13:35 -0500)]
fs/unionfs/: Fix a memory leak in unionfs_read_super

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs: Fix a memory leak & null pointer dereference
Erez Zadok [Fri, 2 Mar 2007 18:00:57 +0000 (13:00 -0500)]
fs/unionfs: Fix a memory leak & null pointer dereference

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years ago[PATCH] UNION_FS must depend on SLAB
Adrian Bunk [Wed, 21 Feb 2007 06:30:10 +0000 (01:30 -0500)]
[PATCH] UNION_FS must depend on SLAB

On Sat, Feb 17, 2007 at 09:51:46PM -0800, Andrew Morton wrote:
>...
> Changes since 2.6.20-mm1:
>...
>  git-unionfs.patch
>...
>  git trees
>...

<--  snip  -->

...
  CC      fs/unionfs/copyup.o
/home/bunk/linux/kernel-2.6/linux-2.6.20-mm2/fs/unionfs/copyup.c: In function 'create_parents_named':
/home/bunk/linux/kernel-2.6/linux-2.6.20-mm2/fs/unionfs/copyup.c:620: error: 'malloc_sizes' undeclared (first use in this function)
/home/bunk/linux/kernel-2.6/linux-2.6.20-mm2/fs/unionfs/copyup.c:620: error: (Each undeclared identifier is reported only once
/home/bunk/linux/kernel-2.6/linux-2.6.20-mm2/fs/unionfs/copyup.c:620: error: for each function it appears in.)
make[3]: *** [fs/unionfs/copyup.o] Error 1

<--  snip  -->

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Remove unused structure members & macros
Josef 'Jeff' Sipek [Sat, 17 Feb 2007 08:25:15 +0000 (03:25 -0500)]
fs/unionfs/: Remove unused structure members & macros

This patch removes:
- hidden_mnt pointer from struct unionfs_data
- mount_flag from struct unionfs_sb_info
- mount_flag related macros

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoeCryptfs: convert lookup_one_len() to lookup_one_len_nd()
Michael Halcrow [Fri, 16 Feb 2007 19:09:25 +0000 (14:09 -0500)]
eCryptfs: convert lookup_one_len() to lookup_one_len_nd()

Call the new lookup_one_len_nd() rather than lookup_one_len().  This fixes an
oops when stacked on NFS.

Note that there are still some issues with eCryptfs on NFS having to do with
directory deletion (I'm not getting an oops, just an -EBUSY).

Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Documentation update
Erez Zadok [Mon, 12 Feb 2007 17:36:38 +0000 (12:36 -0500)]
Unionfs: Documentation update

Be little gentler & updated the URLs

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/: Move eCryptfs & Unionfs config options into a sub-menu
Josef 'Jeff' Sipek [Tue, 20 Feb 2007 04:35:32 +0000 (23:35 -0500)]
fs/: Move eCryptfs & Unionfs config options into a sub-menu

Using The Misc filesystems sub-menu for layered/stackable filesystems only
makes it harder for users to find eCryptfs/Unionfs.

Additionally, the menu can be easily turned into a menuconfig, which could
be used to turn on any VFS/VM functionality required by layered filesystems
(there is none at the moment).

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com>
17 years agofs/unionfs/: Use __roundup_pow_of_two instead of custom rounding code
Josef 'Jeff' Sipek [Thu, 1 Feb 2007 17:14:03 +0000 (12:14 -0500)]
fs/unionfs/: Use __roundup_pow_of_two instead of custom rounding code

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Don't duplicate the struct nameidata
Josef 'Jeff' Sipek [Sun, 28 Jan 2007 19:20:19 +0000 (14:20 -0500)]
fs/unionfs/: Don't duplicate the struct nameidata

The only fields that we have to watch out for are the dentry and vfsmount.
Additionally, this makes Unionfs gentler on the stack as nameidata is rather
large.

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Andrew Morton's comments
Josef 'Jeff' Sipek [Sun, 28 Jan 2007 19:20:49 +0000 (14:20 -0500)]
fs/unionfs/: Andrew Morton's comments

- rename {,un}lock_dentry to unionfs_{,un}lock_dentry
- few minor coding style fixes
- removed prototypes from .c files
- replaced dbstart macros etc with static inlines
- replaced UNIONFS_D(d)->sem semaphore with a mutex
- renamed sioq struct workqueue to superio_workqueue
- made unionfs_get_nlinks and alloc_whname not inlined

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: possible cleanups
Adrian Bunk [Thu, 25 Jan 2007 08:15:59 +0000 (03:15 -0500)]
fs/unionfs/: possible cleanups

This patch contains the following possible cleanups:
- every function should #include the headers containing the prototypes
  of it's global functions
- static functions in C files shouldn't be marked "inline", gcc should
  know best when to inline them
- make needlessly global code static
- #if 0 the following unused global function:
  - stale_inode.c: is_stale_inode()

Signed-off-by: Adrian Bunk <bunk@stusta.de>
[removed stale inode related fixes as stale_inode.c is gone]
Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agofs/unionfs/: Remove stale_inode.c
Josef 'Jeff' Sipek [Sun, 28 Jan 2007 20:52:43 +0000 (15:52 -0500)]
fs/unionfs/: Remove stale_inode.c

The stale inode operations were heavily based on bad inode operations. This
patch removes stale_inode.c and converts all users of stale_inode_ops to
bad_inode_ops as there seems to be no reason to return ESTALE instead of
EIO.

This is the more appropriate than porting the bad_inode.c fix (commit
be6aab0e9fa6d3c6d75aa1e38ac972d8b4ee82b8) to stale_inode.c.

Signed-off-by: Josef 'Jeff' Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Extended Attributes support
Josef "Jeff" Sipek [Sun, 28 Jan 2007 20:11:23 +0000 (15:11 -0500)]
Unionfs: Extended Attributes support

Extended attribute support.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Kconfig and Makefile
Josef "Jeff" Sipek [Sun, 28 Jan 2007 20:10:20 +0000 (15:10 -0500)]
Unionfs: Kconfig and Makefile

This patch contains the changes to fs Kconfig file, Makefiles, and Maintainers
file for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Unlink
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:47:57 +0000 (18:47 -0500)]
Unionfs: Unlink

This patch provides unlink functionality for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Include file
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:47:43 +0000 (18:47 -0500)]
Unionfs: Include file

Global include file - can be included from userspace by utilities.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Internal include file
Josef "Jeff" Sipek [Sun, 28 Jan 2007 20:05:29 +0000 (15:05 -0500)]
Unionfs: Internal include file

This patch contains an internal Unionfs include file. The include file is
specific to kernel code only, and therefore is separate from
include/linux/unionfs.h.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Helper macros/inlines
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:46:56 +0000 (18:46 -0500)]
Unionfs: Helper macros/inlines

This patch contains many macros and inline functions used thoughout Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Handling of stale inodes
Josef "Jeff" Sipek [Sun, 28 Jan 2007 20:05:07 +0000 (15:05 -0500)]
Unionfs: Handling of stale inodes

Provides nicer handling of stale inodes.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Superblock operations
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:46:47 +0000 (18:46 -0500)]
Unionfs: Superblock operations

This patch contains the superblock operations for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Miscellaneous helper functions
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:46:39 +0000 (18:46 -0500)]
Unionfs: Miscellaneous helper functions

This patch contains miscellaneous helper functions used thoughout Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Privileged operations workqueue
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:46:11 +0000 (18:46 -0500)]
Unionfs: Privileged operations workqueue

Workqueue & helper functions used to perform privileged operations on
behalf of the user process.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Rename
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:45:56 +0000 (18:45 -0500)]
Unionfs: Rename

This patch provides rename functionality for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Readdir state
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:45:48 +0000 (18:45 -0500)]
Unionfs: Readdir state

This file contains the routines for maintaining readdir state.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Main module functions
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:45:28 +0000 (18:45 -0500)]
Unionfs: Main module functions

Module init & cleanup code, as well as interposition functions.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Lookup helper functions
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:45:13 +0000 (18:45 -0500)]
Unionfs: Lookup helper functions

This patch provides helper functions for the lookup operations in Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Inode operations
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:44:54 +0000 (18:44 -0500)]
Unionfs: Inode operations

This patch provides the inode operations for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Directory manipulation helper functions
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:44:35 +0000 (18:44 -0500)]
Unionfs: Directory manipulation helper functions

This patch contains directory manipulation helper functions.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Directory file operations
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:44:24 +0000 (18:44 -0500)]
Unionfs: Directory file operations

This patch provides directory file operations.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: File operations
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:44:05 +0000 (18:44 -0500)]
Unionfs: File operations

This patch provides the file operations for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Dentry operations
Josef "Jeff" Sipek [Sun, 28 Jan 2007 20:05:21 +0000 (15:05 -0500)]
Unionfs: Dentry operations

This patch contains the dentry operations for Unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Copyup Functionality
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:42:32 +0000 (18:42 -0500)]
Unionfs: Copyup Functionality

This patch contains the functions used to perform copyup operations in unionfs.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Common file operations
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:42:16 +0000 (18:42 -0500)]
Unionfs: Common file operations

This patch contains helper functions used through the rest of the code which
pertains to files.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoUnionfs: Branch management functionality
Josef "Jeff" Sipek [Sun, 21 Jan 2007 23:42:01 +0000 (18:42 -0500)]
Unionfs: Branch management functionality

This patch contains the ioctls to increase the union generation and to query
which branch a file exists on.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agolookup_one_len_nd - lookup_one_len with nameidata argument
Josef "Jeff" Sipek [Sun, 29 Apr 2007 19:36:03 +0000 (15:36 -0400)]
lookup_one_len_nd - lookup_one_len with nameidata argument

This patch renames lookup_one_len to lookup_one_len_nd, and adds a nameidata
argument. An inline function, lookup_one_len (which calls lookup_one_len_nd
with nd == NULL) preserves original behavior.

The following Unionfs patches depend on this one.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
17 years agoUnionfs: Documentation
Josef "Jeff" Sipek [Tue, 20 Feb 2007 04:31:35 +0000 (23:31 -0500)]
Unionfs: Documentation

This patch contains documentation for Unionfs. You will find several files
outlining basic unification concepts and rename semantics.

Signed-off-by: Josef "Jeff" Sipek <jsipek@cs.sunysb.edu>
Signed-off-by: David Quigley <dquigley@fsl.cs.sunysb.edu>
Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
17 years agoLinux 2.6.23.17 v2.6.23.17
Greg Kroah-Hartman [Tue, 26 Feb 2008 00:14:28 +0000 (16:14 -0800)]
Linux 2.6.23.17

17 years agox86_64: CPA, fix cache attribute inconsistency bug
Ingo Molnar [Fri, 15 Feb 2008 19:58:54 +0000 (20:58 +0100)]
x86_64: CPA, fix cache attribute inconsistency bug

no upstream git id as the code has been rewritten.

fix CPA cache attribute bug in v2.6.23. When phys_base is nonzero
(when CONFIG_RELOCATABLE=y) then change_page_attr_addr() miscalculates
the secondary alias address by -14 MB (depending on the configured
offset).

The default 64-bit kernels of Fedora and Ubuntu are affected:

   $ grep RELOCA /boot/config-2.6.23.9-85.fc8
     CONFIG_RELOCATABLE=y

   $ grep RELOC /boot/config-2.6.22-14-generic
     CONFIG_RELOCATABLE=y

and probably on many other distros as well.

the bug affects all pages in the first 40 MB of physical RAM that
are allocated by some subsystem that does ioremap_nocache() on them:

       if (__pa(address) < KERNEL_TEXT_SIZE) {

Hence we might leave page table entries with inconsistent cache
attributes around (pages mapped at both UnCacheable and Write-Back),
and we can also set the wrong kernel text pages to UnCacheable.

the effects of this bug can be random slowdowns and other misbehavior.
If for example AGP allocates its aperture pages into the first 40 MB
of physical RAM, then the -14 MB bug might mark random kernel texto
pages as uncacheable, slowing down a random portion of the 64-bit
kernel until the AGP driver is unloaded.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoBe more robust about bad arguments in get_user_pages()
Jonathan Corbet [Mon, 11 Feb 2008 23:17:33 +0000 (16:17 -0700)]
Be more robust about bad arguments in get_user_pages()

patch 900cf086fd2fbad07f72f4575449e0d0958f860f in mainline.

So I spent a while pounding my head against my monitor trying to figure
out the vmsplice() vulnerability - how could a failure to check for
*read* access turn into a root exploit? It turns out that it's a buffer
overflow problem which is made easy by the way get_user_pages() is
coded.

In particular, "len" is a signed int, and it is only checked at the
*end* of a do {} while() loop.  So, if it is passed in as zero, the loop
will execute once and decrement len to -1.  At that point, the loop will
proceed until the next invalid address is found; in the process, it will
likely overflow the pages array passed in to get_user_pages().

I think that, if get_user_pages() has been asked to grab zero pages,
that's what it should do.  Thus this patch; it is, among other things,
enough to block the (already fixed) root exploit and any others which
might be lurking in similar code.  I also think that the number of pages
should be unsigned, but changing the prototype of this function probably
requires some more careful review.

Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoDisable G5 NAP mode during SMU commands on U3
Benjamin Herrenschmidt [Thu, 7 Feb 2008 03:29:43 +0000 (14:29 +1100)]
Disable G5 NAP mode during SMU commands on U3

patch 592a607bbc053bc6f614a0e619326009f4b3829e in mainline.

It appears that with the U3 northbridge, if the processor is in NAP
mode the whole time while waiting for an SMU command to complete,
then the SMU will fail.  It could be related to the weird backward
mechanism the SMU uses to get to system memory via i2c to the
northbridge that doesn't operate properly when the said bridge is
in napping along with the CPU.  That is on U3 at least, U4 doesn't
seem to be affected.

This didn't show before NO_HZ as the timer wakeup was enough to make
it work it seems, but that is no longer the case.

This fixes it by disabling NAP mode on those machines while
an SMU command is in flight.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agogenirq: do not leave interupts enabled on free_irq
Thomas Gleixner [Tue, 19 Feb 2008 23:29:02 +0000 (00:29 +0100)]
genirq: do not leave interupts enabled on free_irq

commit 89d694b9dbe769ca1004e01db0ca43964806a611

The default_disable() function was changed in commit:

 76d2160147f43f982dfe881404cfde9fd0a9da21
 genirq: do not mask interrupts by default

It removed the mask function in favour of the default delayed
interrupt disabling. Unfortunately this also broke the shutdown in
free_irq() when the last handler is removed from the interrupt for
those architectures which rely on the default implementations. Now we
can end up with a enabled interrupt line after the last handler was
removed, which can result in spurious interrupts.

Fix this by adding a default_shutdown function, which is only
installed, when the irqchip implementation does provide neither a
shutdown nor a disable function.

Pointed-out-by: Michael Hennerich <Michael.Hennerich@analog.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ingo Molnar <mingo@elte.hu>
Tested-by: Michael Hennerich <Michael.Hennerich@analog.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agohrtimer: check relative timeouts for overflow
Thomas Gleixner [Wed, 20 Feb 2008 00:03:00 +0000 (01:03 +0100)]
hrtimer: check relative timeouts for overflow

commit: 5a7780e725d1bb4c3094fcc12f1c5c5faea1e988

Various user space callers ask for relative timeouts. While we fixed
that overflow issue in hrtimer_start(), the sites which convert
relative user space values to absolute timeouts themself were uncovered.

Instead of putting overflow checks into each place add a function
which does the sanity checking and convert all affected callers to use
it.

Thanks to Frans Pop, who reported the problem and tested the fixes.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ingo Molnar <mingo@elte.hu>
Tested-by: Frans Pop <elendil@planet.nl>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoNETFILTER: nf_conntrack_tcp: conntrack reopening fix
Jozsef Kadlecsik [Tue, 19 Feb 2008 15:24:01 +0000 (16:24 +0100)]
NETFILTER: nf_conntrack_tcp: conntrack reopening fix

[NETFILTER]: nf_conntrack_tcp: conntrack reopening fix

[Upstream commits b2155e7f + d0c1fd7a]

TCP connection tracking in netfilter did not handle TCP reopening
properly: active close was taken into account for one side only and
not for any side, which is fixed now. The patch includes more comments
to explain the logic how the different cases are handled.
The bug was discovered by Jeff Chua.

Signed-off-by: Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoNFS: Fix a potential file corruption issue when writing
Trond Myklebust [Fri, 8 Feb 2008 19:12:11 +0000 (14:12 -0500)]
NFS: Fix a potential file corruption issue when writing

patch 5d47a35600270e7115061cb1320ee60ae9bcb6b8 in mainline.

If the inode is flagged as having an invalid mapping, then we can't rely on
the PageUptodate() flag. Ensure that we don't use the "anti-fragmentation"
write optimisation in nfs_updatepage(), since that will cause NFS to write
out areas of the page that are no longer guaranteed to be up to date.

A potential corruption could occur in the following scenario:

client 1 client 2
=============== ===============
fd=open("f",O_CREAT|O_WRONLY,0644);
write(fd,"fubar\n",6); // cache last page
close(fd);
fd=open("f",O_WRONLY|O_APPEND);
write(fd,"foo\n",4);
close(fd);

fd=open("f",O_WRONLY|O_APPEND);
write(fd,"bar\n",4);
close(fd);
-----
The bug may lead to the file "f" reading 'fubar\n\0\0\0\nbar\n' because
client 2 does not update the cached page after re-opening the file for
write. Instead it keeps it marked as PageUptodate() until someone calls
invaldate_inode_pages2() (typically by calling read()).

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>