[PATCH] make consistent errno value when open pident vs exit(2) race

[PATCH] make consistent errno value when open pident vs exit(2) race

[KOSAKI Motohiro]

proc_pident_instantiate() has following call flow.

proc_pident_lookup()
  proc_pident_instantiate()
    proc_pid_make_inode()

And, proc_pident_lookup() has following error handling.

	const struct pid_entry *p, *last;
	error = ERR_PTR(-ENOENT);
	if (!task)
		goto out_no_task;

Then, proc_pident_instantiate should return ENOENT too when racing against
exit(2) occur.

EINAL has two bad reason. 
  - it implies caller is wrong. bad the race isn't caller's mistake. 
  - man 2 open don't explain EINVAL. user often don't handle it.


Note: Other proc_pid_make_inode() caller already use ENOENT properly.

Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
--
 fs/proc/base.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/proc/base.c b/fs/proc/base.c
index f715597..803d09e 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -1953,7 +1953,7 @@ static struct dentry *proc_pident_instantiate(struct inode *dir,
 	const struct pid_entry *p = ptr;
 	struct inode *inode;
 	struct proc_inode *ei;
-	struct dentry *error = ERR_PTR(-EINVAL);
+	struct dentry *error = ERR_PTR(-ENOENT);
 
 	inode = proc_pid_make_inode(dir->i_sb, task);
 	if (!inode)

Re: [PATCH] make consistent errno value when open pident vs exit(2) race

[Eric W. Biederman]

KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> writes:

> proc_pident_instantiate() has following call flow.
>
> proc_pident_lookup()
>   proc_pident_instantiate()
>     proc_pid_make_inode()
>
> And, proc_pident_lookup() has following error handling.
>
> 	const struct pid_entry *p, *last;
> 	error = ERR_PTR(-ENOENT);
> 	if (!task)
> 		goto out_no_task;
>
> Then, proc_pident_instantiate should return ENOENT too when racing against
> exit(2) occur.
>
> EINAL has two bad reason. 
>   - it implies caller is wrong. bad the race isn't caller's mistake. 
>   - man 2 open don't explain EINVAL. user often don't handle it.
>
>
> Note: Other proc_pid_make_inode() caller already use ENOENT properly.

Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>

Looks reasonable to me.


> Cc: Eric W. Biederman <ebiederm@xmission.com>
> Cc: Andrew Morton <akpm@osdl.org>
> Cc: Alexey Dobriyan <adobriyan@gmail.com>
> Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
> --
>  fs/proc/base.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/fs/proc/base.c b/fs/proc/base.c
> index f715597..803d09e 100644
> --- a/fs/proc/base.c
> +++ b/fs/proc/base.c
> @@ -1953,7 +1953,7 @@ static struct dentry *proc_pident_instantiate(struct inode *dir,
>  	const struct pid_entry *p = ptr;
>  	struct inode *inode;
>  	struct proc_inode *ei;
> -	struct dentry *error = ERR_PTR(-EINVAL);
> +	struct dentry *error = ERR_PTR(-ENOENT);
>  
>  	inode = proc_pid_make_inode(dir->i_sb, task);
>  	if (!inode)