TAKE 981875 - xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()

TAKE 981875 - xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()

[Lachlan McIlroy]

xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()

We need a vfsmount pointer in xfs_dm_rdwr() but we are not
provided with one and there's no way to get to it.  So add
a m_vfsmount field to the xfs_mount structure and set it
up at mount time.  We can then access it through the xfs
inode.

Date:  Fri May 16 14:56:46 AEST 2008
Workarea:  redback.melbourne.sgi.com:/home/lachlan/isms/2.6.x-bmap
Inspected by:  dgc
Author:  lachlan

The following file(s) were checked into:
  longdrop.melbourne.sgi.com:/isms/linux/2.6.x-xfs-melb


Modid:  xfs-linux-melb:xfs-kern:31176a
fs/xfs/xfs_mount.h - 1.266 - changed
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-linux/xfs_mount.h.diff?r1=text&tr1=1.266&r2=text&tr2=1.265&f=h
fs/xfs/linux-2.6/xfs_super.c - 1.417 - changed
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-linux/linux-2.6/xfs_super.c.diff?r1=text&tr1=1.417&r2=text&tr2=1.416&f=h
	- Set up a vfsmount pointer in the xfs mount structure.

fs/xfs/dmapi/xfs_dm.c - 1.73 - changed
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-linux/dmapi/xfs_dm.c.diff?r1=text&tr1=1.73&r2=text&tr2=1.72&f=h
	- xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()

Re: TAKE 981875 - xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()

[Christoph Hellwig]

On Fri, May 16, 2008 at 02:57:32PM +1000, Lachlan McIlroy wrote:
> xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()
> 
> We need a vfsmount pointer in xfs_dm_rdwr() but we are not
> provided with one and there's no way to get to it.  So add
> a m_vfsmount field to the xfs_mount structure and set it
> up at mount time.  We can then access it through the xfs
> inode.

Stong NACK for this crap, and double NACK for pushing this towards
mainline.

Re: TAKE 981875 - xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()

[Christoph Hellwig]

On Fri, May 16, 2008 at 01:10:48AM -0400, Christoph Hellwig wrote:
> On Fri, May 16, 2008 at 02:57:32PM +1000, Lachlan McIlroy wrote:
> > xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()
> > 
> > We need a vfsmount pointer in xfs_dm_rdwr() but we are not
> > provided with one and there's no way to get to it.  So add
> > a m_vfsmount field to the xfs_mount structure and set it
> > up at mount time.  We can then access it through the xfs
> > inode.
> 
> Stong NACK for this crap, and double NACK for pushing this towards
> mainline.

No reply yet.  Folks, you realize this is an instant oops if you perform
dmapi operations after one fs instance has been unmounted?

Re: TAKE 981875 - xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()

[Niv Sardi]

Christoph Hellwig <hch@infradead.org> writes:

> On Fri, May 16, 2008 at 01:10:48AM -0400, Christoph Hellwig wrote:
>> On Fri, May 16, 2008 at 02:57:32PM +1000, Lachlan McIlroy wrote:
>> > xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()
>> > 
>> > We need a vfsmount pointer in xfs_dm_rdwr() but we are not
>> > provided with one and there's no way to get to it.  So add
>> > a m_vfsmount field to the xfs_mount structure and set it
>> > up at mount time.  We can then access it through the xfs
>> > inode.
>> 
>> Stong NACK for this crap, and double NACK for pushing this towards
>> mainline.
>
> No reply yet.  Folks, you realize this is an instant oops if you perform
> dmapi operations after one fs instance has been unmounted?

This is not going into mainline, but we need it in -dev to make sure we
can still QA DMAPI. We (I should be but flooded with other stuff) are
looking at other solutions to fix the issue the right way.

Cheers,
-- 
Niv Sardi

Re: TAKE 981875 - xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()

[Christoph Hellwig]

Not only is this buggy stuff still there, but it also sneaked into
linux-next were it shold be reverted ASAP.

Re: TAKE 981875 - xfs_dm_rdwr() needs to pass a vfsmount to dentry_open()

[Niv Sardi]

Christoph Hellwig <hch@infradead.org> writes:

> Not only is this buggy stuff still there, but it also sneaked into
> linux-next were it shold be reverted ASAP.

Yep, shouldn't have gotten there, Lachlan and I are working on cleaning
the mess in the oss git tree, the pull should be made out of this one:


Cheers,
-- 
Niv Sardi