* [PATCH] NFS: Don't hard-code the fs_type when submounting
@ 2020-02-20 13:06 Scott Mayhew
2020-02-20 13:46 ` Patrick Steinhardt
0 siblings, 1 reply; 4+ messages in thread
From: Scott Mayhew @ 2020-02-20 13:06 UTC (permalink / raw)
To: trond.myklebust, anna.schumaker; +Cc: ps, dhowells, linux-nfs
Hard-coding the fstype causes "nfs4" mounts to appear as "nfs",
which breaks scripts that do "umount -at nfs4".
Reported-by: Patrick Steinhardt <ps@pks.im>
Fixes: f2aedb713c28 ("NFS: Add fs_context support.")
Signed-off-by: Scott Mayhew <smayhew@redhat.com>
---
fs/nfs/namespace.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c
index ad6077404947..f3ece8ed3203 100644
--- a/fs/nfs/namespace.c
+++ b/fs/nfs/namespace.c
@@ -153,7 +153,7 @@ struct vfsmount *nfs_d_automount(struct path *path)
/* Open a new filesystem context, transferring parameters from the
* parent superblock, including the network namespace.
*/
- fc = fs_context_for_submount(&nfs_fs_type, path->dentry);
+ fc = fs_context_for_submount(path->mnt->mnt_sb->s_type, path->dentry);
if (IS_ERR(fc))
return ERR_CAST(fc);
--
2.24.1
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] NFS: Don't hard-code the fs_type when submounting
2020-02-20 13:06 [PATCH] NFS: Don't hard-code the fs_type when submounting Scott Mayhew
@ 2020-02-20 13:46 ` Patrick Steinhardt
2020-02-21 20:21 ` Scott Mayhew
0 siblings, 1 reply; 4+ messages in thread
From: Patrick Steinhardt @ 2020-02-20 13:46 UTC (permalink / raw)
To: Scott Mayhew; +Cc: trond.myklebust, anna.schumaker, dhowells, linux-nfs
[-- Attachment #1: Type: text/plain, Size: 1452 bytes --]
On Thu, Feb 20, 2020 at 08:06:20AM -0500, Scott Mayhew wrote:
> Hard-coding the fstype causes "nfs4" mounts to appear as "nfs",
> which breaks scripts that do "umount -at nfs4".
>
> Reported-by: Patrick Steinhardt <ps@pks.im>
> Fixes: f2aedb713c28 ("NFS: Add fs_context support.")
> Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> ---
> fs/nfs/namespace.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c
> index ad6077404947..f3ece8ed3203 100644
> --- a/fs/nfs/namespace.c
> +++ b/fs/nfs/namespace.c
> @@ -153,7 +153,7 @@ struct vfsmount *nfs_d_automount(struct path *path)
> /* Open a new filesystem context, transferring parameters from the
> * parent superblock, including the network namespace.
> */
> - fc = fs_context_for_submount(&nfs_fs_type, path->dentry);
> + fc = fs_context_for_submount(path->mnt->mnt_sb->s_type, path->dentry);
> if (IS_ERR(fc))
> return ERR_CAST(fc);
Thanks for your fix! While this fixes the fstype with mount.nfs4(8),
it still doesn't work when using mount(8):
$ sudo mount server:/mnt /mnt && findmnt -n -ofstype /mnt
nfs
$ sudo umount /mnt
$ sudo mount.nfs4 server:/mnt /mnt && findmnt -n -ofstype /mnt
nfs4
I guess the issue is that the kernel doesn't yet know which NFS version
the server provides at the point where `fs_context_for_submount()` is
called.
Patrick
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] NFS: Don't hard-code the fs_type when submounting
2020-02-20 13:46 ` Patrick Steinhardt
@ 2020-02-21 20:21 ` Scott Mayhew
2020-02-22 7:25 ` Patrick Steinhardt
0 siblings, 1 reply; 4+ messages in thread
From: Scott Mayhew @ 2020-02-21 20:21 UTC (permalink / raw)
To: Patrick Steinhardt; +Cc: trond.myklebust, anna.schumaker, dhowells, linux-nfs
On Thu, 20 Feb 2020, Patrick Steinhardt wrote:
> On Thu, Feb 20, 2020 at 08:06:20AM -0500, Scott Mayhew wrote:
> > Hard-coding the fstype causes "nfs4" mounts to appear as "nfs",
> > which breaks scripts that do "umount -at nfs4".
> >
> > Reported-by: Patrick Steinhardt <ps@pks.im>
> > Fixes: f2aedb713c28 ("NFS: Add fs_context support.")
> > Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> > ---
> > fs/nfs/namespace.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c
> > index ad6077404947..f3ece8ed3203 100644
> > --- a/fs/nfs/namespace.c
> > +++ b/fs/nfs/namespace.c
> > @@ -153,7 +153,7 @@ struct vfsmount *nfs_d_automount(struct path *path)
> > /* Open a new filesystem context, transferring parameters from the
> > * parent superblock, including the network namespace.
> > */
> > - fc = fs_context_for_submount(&nfs_fs_type, path->dentry);
> > + fc = fs_context_for_submount(path->mnt->mnt_sb->s_type, path->dentry);
> > if (IS_ERR(fc))
> > return ERR_CAST(fc);
>
> Thanks for your fix! While this fixes the fstype with mount.nfs4(8),
> it still doesn't work when using mount(8):
>
> $ sudo mount server:/mnt /mnt && findmnt -n -ofstype /mnt
> nfs
> $ sudo umount /mnt
> $ sudo mount.nfs4 server:/mnt /mnt && findmnt -n -ofstype /mnt
> nfs4
>
> I guess the issue is that the kernel doesn't yet know which NFS version
> the server provides at the point where `fs_context_for_submount()` is
> called.
Thanks for testing. Actually the problem is that the super_block's
s_type is now based on the fs_context->fs_type field, which is set when
the fs_context is created (way before the mount options are parsed).
When you use mount(8) without specifying '-t nfs4', it defaults to
using the mount.nfs helper, which calls mount(2) with 'nfs' as the
fstype. I'm sending a second patch that double-checks the
fs_context->fs_type after the mount options have been parsed. We still
shouldn't be hard-coding the fstype in fs_context_for_submount() though
(i.e. both patches are needed).
-Scott
>
> Patrick
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] NFS: Don't hard-code the fs_type when submounting
2020-02-21 20:21 ` Scott Mayhew
@ 2020-02-22 7:25 ` Patrick Steinhardt
0 siblings, 0 replies; 4+ messages in thread
From: Patrick Steinhardt @ 2020-02-22 7:25 UTC (permalink / raw)
To: Scott Mayhew; +Cc: trond.myklebust, anna.schumaker, dhowells, linux-nfs
[-- Attachment #1: Type: text/plain, Size: 2407 bytes --]
On Fri, Feb 21, 2020 at 03:21:05PM -0500, Scott Mayhew wrote:
> On Thu, 20 Feb 2020, Patrick Steinhardt wrote:
>
> > On Thu, Feb 20, 2020 at 08:06:20AM -0500, Scott Mayhew wrote:
> > > Hard-coding the fstype causes "nfs4" mounts to appear as "nfs",
> > > which breaks scripts that do "umount -at nfs4".
> > >
> > > Reported-by: Patrick Steinhardt <ps@pks.im>
> > > Fixes: f2aedb713c28 ("NFS: Add fs_context support.")
> > > Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> > > ---
> > > fs/nfs/namespace.c | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c
> > > index ad6077404947..f3ece8ed3203 100644
> > > --- a/fs/nfs/namespace.c
> > > +++ b/fs/nfs/namespace.c
> > > @@ -153,7 +153,7 @@ struct vfsmount *nfs_d_automount(struct path *path)
> > > /* Open a new filesystem context, transferring parameters from the
> > > * parent superblock, including the network namespace.
> > > */
> > > - fc = fs_context_for_submount(&nfs_fs_type, path->dentry);
> > > + fc = fs_context_for_submount(path->mnt->mnt_sb->s_type, path->dentry);
> > > if (IS_ERR(fc))
> > > return ERR_CAST(fc);
> >
> > Thanks for your fix! While this fixes the fstype with mount.nfs4(8),
> > it still doesn't work when using mount(8):
> >
> > $ sudo mount server:/mnt /mnt && findmnt -n -ofstype /mnt
> > nfs
> > $ sudo umount /mnt
> > $ sudo mount.nfs4 server:/mnt /mnt && findmnt -n -ofstype /mnt
> > nfs4
> >
> > I guess the issue is that the kernel doesn't yet know which NFS version
> > the server provides at the point where `fs_context_for_submount()` is
> > called.
>
> Thanks for testing. Actually the problem is that the super_block's
> s_type is now based on the fs_context->fs_type field, which is set when
> the fs_context is created (way before the mount options are parsed).
> When you use mount(8) without specifying '-t nfs4', it defaults to
> using the mount.nfs helper, which calls mount(2) with 'nfs' as the
> fstype. I'm sending a second patch that double-checks the
> fs_context->fs_type after the mount options have been parsed. We still
> shouldn't be hard-coding the fstype in fs_context_for_submount() though
> (i.e. both patches are needed).
I can confirm that the problem is fixed with both patches applied.
Thanks a lot!
Patrick
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2020-02-22 7:25 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-02-20 13:06 [PATCH] NFS: Don't hard-code the fs_type when submounting Scott Mayhew
2020-02-20 13:46 ` Patrick Steinhardt
2020-02-21 20:21 ` Scott Mayhew
2020-02-22 7:25 ` Patrick Steinhardt
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.