proc: mounts_poll() make consistent to mdstat_poll
authorKOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Thu, 9 Apr 2009 04:57:59 +0000 (13:57 +0900)
committerGreg Kroah-Hartman <gregkh@suse.de>
Thu, 16 Apr 2009 23:17:10 +0000 (16:17 -0700)
In recently sysfs_poll discussion, Neil Brown pointed out /proc/mounts
also should be fixed.

SUSv3 says "Regular files shall always poll TRUE for reading and
writing".  see
http://www.opengroup.org/onlinepubs/009695399/functions/poll.html

Then, mounts_poll()'s default should be "POLLIN | POLLRDNORM".  it mean
always readable.

In addition, event trigger should use "POLLERR | POLLPRI" instead
POLLERR.  it makes consistent to mdstat_poll() and sysfs_poll(). and,
select(2) can handle POLLPRI easily.

Reported-by: Neil Brown <neilb@suse.de>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Ram Pai <linuxram@us.ibm.com>
Cc: Miklos Szeredi <mszeredi@suse.cz>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
fs/proc/base.c

index f71559784bfb9bbc70b4d021cd2120c4bfa2121c..aa763ab007775f7c7da2a477e8df148dfad5f940 100644 (file)
@@ -648,14 +648,14 @@ static unsigned mounts_poll(struct file *file, poll_table *wait)
 {
        struct proc_mounts *p = file->private_data;
        struct mnt_namespace *ns = p->ns;
-       unsigned res = 0;
+       unsigned res = POLLIN | POLLRDNORM;
 
        poll_wait(file, &ns->poll, wait);
 
        spin_lock(&vfsmount_lock);
        if (p->event != ns->event) {
                p->event = ns->event;
-               res = POLLERR;
+               res |= POLLERR | POLLPRI;
        }
        spin_unlock(&vfsmount_lock);