Re: [patch] remove BKL from inode_setattr

[Steve Lord <lord@sgi.com>]

On Mon, 2002-10-14 at 01:34, Andrew Morton wrote:

> 
> The number of filsystems which do not take the bkl in truncate/setattr
> is in fact quite small.  Here's the patch which removes all doubt:
> 
> 
> 
> 
>  fs/affs/file.c          |   13 ++++++++-----
>  fs/attr.c               |    2 --
>  fs/cifs/inode.c         |    7 ++++++-
>  fs/jfs/file.c           |    3 +++
>  fs/reiserfs/file.c      |    2 ++
>  fs/smbfs/proc.c         |   18 +++++++++++++++---
>  fs/sysv/itree.c         |    6 +++++-
>  fs/xfs/linux/xfs_iops.c |   11 +++++++++--
>  8 files changed, 48 insertions(+), 14 deletions(-)

XFS deliberately does not take the BKL - anywhere. Our setattr
code is doing its own locking. You just added the BKL to a 
bunch of xfs operations which do not need it. Now, vmtruncate
may need it, itself, but if vmtruncate does not, then the xfs
callout from vmtruncate certainly does not.

Steve

-- 

Steve Lord                                      voice: +1-651-683-3511
Principal Engineer, Filesystem Software         email: lord@sgi.com