Re: [patch] mm, hugetlb: add thread name and pid to SHM_HUGETLBmlock rlimit warning

[David Rientjes]

On Thu, 8 Mar 2012, Andrew Morton wrote:

> > --- a/fs/hugetlbfs/inode.c
> > +++ b/fs/hugetlbfs/inode.c
> > @@ -946,7 +946,11 @@ struct file *hugetlb_file_setup(const char *name, size_t size,
> >  	if (creat_flags == HUGETLB_SHMFS_INODE && !can_do_hugetlb_shm()) {
> >  		*user = current_user();
> >  		if (user_shm_lock(size, *user)) {
> > -			printk_once(KERN_WARNING "Using mlock ulimits for SHM_HUGETLB is deprecated\n");
> > +			task_lock(current);
> > +			printk_once(KERN_WARNING
> > +				"%s (%d): Using mlock ulimits for SHM_HUGETLB is deprecated\n",
> > +				current->comm, current->pid);
> > +			task_unlock(current);
> 
> I assume the task_lock() is there to protect current->comm.

Yup.

> If so, it
> is unneeded - we're protecting against prctl(PR_SET_NAME), and
> PR_SET_NAME only operates on current, and we know this task isn't
> currently running PR_SET_NAME.
> 
> If there's a way for another task to alter this task's ->comm then we
> _do_ need locking.  But there isn't a way, I hope.
> 

I wish there wasn't as well, it would prevent a lot of the currently buggy 
reads to current->comm and allow us to avoid so many otherwise pointless 
task_lock()s.

This protects against /proc/pid/comm, which is writable by threads in the 
same thread group.  We have a get_task_comm() that does the task_lock() 
internally but requires a TASK_COMM_LEN buffer in the calling code.  It's 
just easier for the calling code to the task_lock() itself for a tiny 
little printk().
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/