[PATCH] procfs: fix tid fdinfo

[PATCH] procfs: fix tid fdinfo

[Jerome Marchand]

Correct the file_operations struct in fdinfo entry of tid_base_stuff[].

Signed-off-by: Jerome Marchand <jmarchan@redhat.com>
---
 base.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/proc/base.c b/fs/proc/base.c
index 7621db8..8418fcc 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -2909,7 +2909,7 @@ out_no_task:
  */
 static const struct pid_entry tid_base_stuff[] = {
 	DIR("fd",        S_IRUSR|S_IXUSR, proc_fd_inode_operations, proc_fd_operations),
-	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fd_operations),
+	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fdinfo_operations),
 	REG("environ",   S_IRUSR, proc_environ_operations),
 	INF("auxv",      S_IRUSR, proc_pid_auxv),
 	ONE("status",    S_IRUGO, proc_pid_status),

Re: [PATCH] procfs: fix tid fdinfo

[Andrew Morton]

On Fri, 23 Apr 2010 17:40:07 +0200
Jerome Marchand <jmarchan@redhat.com> wrote:

> Correct the file_operations struct in fdinfo entry of tid_base_stuff[].
> 
> Signed-off-by: Jerome Marchand <jmarchan@redhat.com>
> ---
>  base.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/fs/proc/base.c b/fs/proc/base.c
> index 7621db8..8418fcc 100644
> --- a/fs/proc/base.c
> +++ b/fs/proc/base.c
> @@ -2909,7 +2909,7 @@ out_no_task:
>   */
>  static const struct pid_entry tid_base_stuff[] = {
>  	DIR("fd",        S_IRUSR|S_IXUSR, proc_fd_inode_operations, proc_fd_operations),
> -	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fd_operations),
> +	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fdinfo_operations),
>  	REG("environ",   S_IRUSR, proc_environ_operations),
>  	INF("auxv",      S_IRUSR, proc_pid_auxv),
>  	ONE("status",    S_IRUGO, proc_pid_status),

hm, the code's been like that for over a year.  What are the
user-visible effects of the bug, and of this change?

Re: [PATCH] procfs: fix tid fdinfo

[Miklos Szeredi]

On Sat, 2010-04-24 at 22:27 +0300, Alexey Dobriyan wrote:
> On Fri, Apr 23, 2010 at 11:16:20AM -0700, Andrew Morton wrote:
> > On Fri, 23 Apr 2010 17:40:07 +0200
> > > -	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fd_operations),
> > > +	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fdinfo_operations),
> 
> > hm, the code's been like that for over a year.  What are the
> > user-visible effects of the bug, and of this change?
> 
> /proc/*/task/*/fdinfo contains symlinks to opened files like /proc/*/fd/
> which is cool bug :-)

Yeah, at least now we can see that someone has started using this
interface ;)

Jerome, thanks for spotting this!

Miklos

Re: [PATCH] procfs: fix tid fdinfo

[Alexey Dobriyan]

On Fri, Apr 23, 2010 at 05:40:07PM +0200, Jerome Marchand wrote:
> -	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fd_operations),
> +	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fdinfo_operations),

Ick.

this is for stable@ as well

Cc: stable@kernel.org

Re: [PATCH] procfs: fix tid fdinfo

[Alexey Dobriyan]

On Fri, Apr 23, 2010 at 11:16:20AM -0700, Andrew Morton wrote:
> On Fri, 23 Apr 2010 17:40:07 +0200
> > -	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fd_operations),
> > +	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fdinfo_operations),

> hm, the code's been like that for over a year.  What are the
> user-visible effects of the bug, and of this change?

/proc/*/task/*/fdinfo contains symlinks to opened files like /proc/*/fd/
which is cool bug :-)

Re: [PATCH] procfs: fix tid fdinfo

[Jerome Marchand]

On 04/23/2010 10:07 PM, Miklos Szeredi wrote:
> On Sat, 2010-04-24 at 22:27 +0300, Alexey Dobriyan wrote:
>> On Fri, Apr 23, 2010 at 11:16:20AM -0700, Andrew Morton wrote:
>>> On Fri, 23 Apr 2010 17:40:07 +0200
>>>> -	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fd_operations),
>>>> +	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fdinfo_operations),
>>
>>> hm, the code's been like that for over a year.  What are the
>>> user-visible effects of the bug, and of this change?
>>
>> /proc/*/task/*/fdinfo contains symlinks to opened files like /proc/*/fd/
>> which is cool bug :-)
> 
> Yeah, at least now we can see that someone has started using this
> interface ;)

AFAIK there is still nobody using it, but lsof should use it, and 
/proc/<pid>/task/<tid>/fd/ too. I haven't checked the code, but as it is
now, it doesn't spot files open by a thread which does share its father's
file descriptor table.

> 
> Jerome, thanks for spotting this!
> 
> Miklos
> 
> 

Re: [PATCH] procfs: fix tid fdinfo

[Miklos Szeredi]

On Mon, 2010-04-26 at 09:49 +0200, Jerome Marchand wrote:
> On 04/23/2010 10:07 PM, Miklos Szeredi wrote:
> > On Sat, 2010-04-24 at 22:27 +0300, Alexey Dobriyan wrote:
> >> On Fri, Apr 23, 2010 at 11:16:20AM -0700, Andrew Morton wrote:
> >>> On Fri, 23 Apr 2010 17:40:07 +0200
> >>>> -	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fd_operations),
> >>>> +	DIR("fdinfo",    S_IRUSR|S_IXUSR, proc_fdinfo_inode_operations, proc_fdinfo_operations),
> >>
> >>> hm, the code's been like that for over a year.  What are the
> >>> user-visible effects of the bug, and of this change?
> >>
> >> /proc/*/task/*/fdinfo contains symlinks to opened files like /proc/*/fd/
> >> which is cool bug :-)
> > 
> > Yeah, at least now we can see that someone has started using this
> > interface ;)
> 
> AFAIK there is still nobody using it, but lsof should use it, and 
> /proc/<pid>/task/<tid>/fd/ too. I haven't checked the code, but as it is
> now, it doesn't spot files open by a thread which does share its father's
> file descriptor table.

Yes, apparently "lsof" doesn't care about contents of /proc/$PID/task.
Well, tasks with separate file descriptor tables are rare, POSIX threads
share their descriptor tables.  But it would be nice if this were fixed.

Thanks,
Miklos