[PATCH] autofs4-2.6.34-rc1 - fix link_count usage

[PATCH] autofs4-2.6.34-rc1 - fix link_count usage

[Ian Kent]

Following the commit below in 2.6.34-rc1 autofs direct mounts stop
working. This is caused by current->link_count being 0 when
->follow_link() is called from do_filp_open().

I can't work out why this hasn't been seen before Als patch series.

This patch removes the autofs dependence on current->link_count.

1f36f774b22a0ceb7dd33eca626746c81a97b6a5 is the first bad commit
commit 1f36f774b22a0ceb7dd33eca626746c81a97b6a5
Author: Al Viro <viro@zeniv.linux.org.uk>
Date:   Sat Dec 26 10:56:19 2009 -0500

    Switch !O_CREAT case to use of do_last()

    ... and now we have all intents crap well localized

    Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

:040000 040000 53c1effc5b22746bb83cdbc6a419bf898067882d 9606f7275fde188a056fd174aa1d622ac0630893 M	fs
---

 fs/autofs4/root.c |    5 ++---
 1 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/fs/autofs4/root.c b/fs/autofs4/root.c
index 109a6c6..a7e04ad 100644
--- a/fs/autofs4/root.c
+++ b/fs/autofs4/root.c
@@ -177,8 +177,7 @@ static int try_to_fill_dentry(struct dentry *dentry, int flags)
 		}
 	/* Trigger mount for path component or follow link */
 	} else if (ino->flags & AUTOFS_INF_PENDING ||
-			autofs4_need_mount(flags) ||
-			current->link_count) {
+			autofs4_need_mount(flags)) {
 		DPRINTK("waiting for mount name=%.*s",
 			dentry->d_name.len, dentry->d_name.name);
 
@@ -262,7 +261,7 @@ static void *autofs4_follow_link(struct dentry *dentry, struct nameidata *nd)
 		spin_unlock(&dcache_lock);
 		spin_unlock(&sbi->fs_lock);
 
-		status = try_to_fill_dentry(dentry, 0);
+		status = try_to_fill_dentry(dentry, lookup_type);
 		if (status)
 			goto out_error;
 

Re: [PATCH] autofs4-2.6.34-rc1 - fix link_count usage

[Ian Kent]

On Mon, 2010-05-10 at 13:18 +0800, Ian Kent wrote:

Sorry, this isn't right, please ignore it.
Apart from the lack of a Signed-off-by: I hadn't tested against
the current kernel source when I posted.

> Following the commit below in 2.6.34-rc1 autofs direct mounts stop
> working. This is caused by current->link_count being 0 when
> ->follow_link() is called from do_filp_open().
> 
> I can't work out why this hasn't been seen before Als patch series.
> 
> This patch removes the autofs dependence on current->link_count.
> 
> 1f36f774b22a0ceb7dd33eca626746c81a97b6a5 is the first bad commit
> commit 1f36f774b22a0ceb7dd33eca626746c81a97b6a5
> Author: Al Viro <viro@zeniv.linux.org.uk>
> Date:   Sat Dec 26 10:56:19 2009 -0500
> 
>     Switch !O_CREAT case to use of do_last()
> 
>     ... and now we have all intents crap well localized
> 
>     Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
> 
> :040000 040000 53c1effc5b22746bb83cdbc6a419bf898067882d 9606f7275fde188a056fd174aa1d622ac0630893 M	fs
> ---
> 
>  fs/autofs4/root.c |    5 ++---
>  1 files changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/fs/autofs4/root.c b/fs/autofs4/root.c
> index 109a6c6..a7e04ad 100644
> --- a/fs/autofs4/root.c
> +++ b/fs/autofs4/root.c
> @@ -177,8 +177,7 @@ static int try_to_fill_dentry(struct dentry *dentry, int flags)
>  		}
>  	/* Trigger mount for path component or follow link */
>  	} else if (ino->flags & AUTOFS_INF_PENDING ||
> -			autofs4_need_mount(flags) ||
> -			current->link_count) {
> +			autofs4_need_mount(flags)) {
>  		DPRINTK("waiting for mount name=%.*s",
>  			dentry->d_name.len, dentry->d_name.name);
>  
> @@ -262,7 +261,7 @@ static void *autofs4_follow_link(struct dentry *dentry, struct nameidata *nd)
>  		spin_unlock(&dcache_lock);
>  		spin_unlock(&sbi->fs_lock);
>  
> -		status = try_to_fill_dentry(dentry, 0);
> +		status = try_to_fill_dentry(dentry, lookup_type);
>  		if (status)
>  			goto out_error;
>  
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

[PATCH] autofs4-2.6.34-rc1 - fix link_count usage

[Ian Kent]

Following the commit below in 2.6.34-rc1 autofs direct mounts stop
working. This is caused by current->link_count being 0 when
->follow_link() is called from do_filp_open().

I can't work out why this hasn't been seen before Als patch series.

This patch removes the autofs dependence on current->link_count.

1f36f774b22a0ceb7dd33eca626746c81a97b6a5 is the first bad commit
commit 1f36f774b22a0ceb7dd33eca626746c81a97b6a5
Author: Al Viro <viro@zeniv.linux.org.uk>
Date:   Sat Dec 26 10:56:19 2009 -0500

    Switch !O_CREAT case to use of do_last()

    ... and now we have all intents crap well localized

    Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

:040000 040000 53c1effc5b22746bb83cdbc6a419bf898067882d 9606f7275fde188a056fd174aa1d622ac0630893 M	fs

Signed-off-by: Ian Kent <raven@themaw.net>
---

 fs/autofs4/root.c |    5 ++---
 1 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/fs/autofs4/root.c b/fs/autofs4/root.c
index 109a6c6..e8e5e63 100644
--- a/fs/autofs4/root.c
+++ b/fs/autofs4/root.c
@@ -177,8 +177,7 @@ static int try_to_fill_dentry(struct dentry *dentry, int flags)
 		}
 	/* Trigger mount for path component or follow link */
 	} else if (ino->flags & AUTOFS_INF_PENDING ||
-			autofs4_need_mount(flags) ||
-			current->link_count) {
+			autofs4_need_mount(flags)) {
 		DPRINTK("waiting for mount name=%.*s",
 			dentry->d_name.len, dentry->d_name.name);
 
@@ -262,7 +261,7 @@ static void *autofs4_follow_link(struct dentry *dentry, struct nameidata *nd)
 		spin_unlock(&dcache_lock);
 		spin_unlock(&sbi->fs_lock);
 
-		status = try_to_fill_dentry(dentry, 0);
+		status = try_to_fill_dentry(dentry, nd->flags);
 		if (status)
 			goto out_error;