Re: [PATCH 3/3][mmotm] updateing size of kcore

[KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>]

On Thu, 17 Sep 2009 14:59:35 +0800
AmA(C)rico Wang <xiyou.wangcong@gmail.com> wrote:

> On Thu, Sep 17, 2009 at 10:45 AM, KAMEZAWA Hiroyuki
> <kamezawa.hiroyu@jp.fujitsu.com> wrote:
> >
> > After memory hotplug (or other events in future), kcore size
> > can be modified.
> >
> > To update inode->i_size, we have to know inode/dentry but we
> > can't get it from inside /proc directly.
> > But considerinyg memory hotplug, kcore image is updated only when
> > it's opened. Then, updating inode->i_size at open() is enough.
> >
> > Cc: WANG Cong <xiyou.wangcong@gmail.com>
> > Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> 
> 
> This patch looks fine.
> 
> However, I am thinking if kcore is the only file under /proc whose size
> is changed dynamically? If no, that probably means we need to change
> generic proc code.
> 
I tried yesteray, and back to this ;)

One thing which makes me confused is that there is no way to get
inode or dentry from proc_dir_entry.

I tried to rewrite proc_getattr() for cheating "ls". But it just works for
"stat" and inode->i_size is used more widely. So, this implementation now.

But considering practically, inode->i_size itself is not meaningful in /proc
files even if it's correct. For example, /proc/vmstat or /proc/stat,
/proc/<pid>/maps... etc...

inode->i_size will be dynamically changed while reading. Now, most of users
know regular files under /proc is not a "real" file and handle them in proper
way. (programs can be used with pipe/stdin works well.)

I wonder /proc/kcore is a special one, which gdb/objdump/readelf may access.
Above 3 programs are for "usual" files and not considering pseudo files under
/proc. So, I think adding generic i->i_size support is an overkill until
there are users depends on that.

Thanks,
-Kame


> Thanks!
> 
> > ---
> > A fs/proc/kcore.c | A  A 5 +++++
> > A 1 file changed, 5 insertions(+)
> >
> > Index: mmotm-2.6.31-Sep14/fs/proc/kcore.c
> > ===================================================================
> > --- mmotm-2.6.31-Sep14.orig/fs/proc/kcore.c
> > +++ mmotm-2.6.31-Sep14/fs/proc/kcore.c
> > @@ -546,6 +546,11 @@ static int open_kcore(struct inode *inod
> > A  A  A  A  A  A  A  A return -EPERM;
> > A  A  A  A if (kcore_need_update)
> > A  A  A  A  A  A  A  A kcore_update_ram();
> > + A  A  A  if (i_size_read(inode) != proc_root_kcore->size) {
> > + A  A  A  A  A  A  A  mutex_lock(&inode->i_mutex);
> > + A  A  A  A  A  A  A  i_size_write(inode, proc_root_kcore->size);
> > + A  A  A  A  A  A  A  mutex_unlock(&inode->i_mutex);
> > + A  A  A  }
> > A  A  A  A return 0;
> > A }
> >
> >
> >
> 

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>