Re: [PATCH -next v2 10/26] tty: Don't take tty_mutex for tty count changes

[Peter Hurley <peter@hurleysoftware.com>]

On 11/05/2014 09:39 PM, Peter Hurley wrote:
> On 11/05/2014 09:33 PM, Greg Kroah-Hartman wrote:
>> On Wed, Nov 05, 2014 at 12:12:53PM -0500, Peter Hurley wrote:
>>> Holding tty_mutex is no longer required to serialize changes to
>>> the tty_count or to prevent concurrent opens of closing ttys;
>>> tty_lock() is sufficient.
>>>
>>> Reviewed-by: Alan Cox <alan@linux.intel.com>
>>> Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
>>> ---
>>>  drivers/tty/tty_io.c | 6 ------
>>>  1 file changed, 6 deletions(-)
>>>
>>> diff --git a/drivers/tty/tty_io.c b/drivers/tty/tty_io.c
>>> index ea8c6cae8d12..e59de81c39a9 100644
>>> --- a/drivers/tty/tty_io.c
>>> +++ b/drivers/tty/tty_io.c
>>> @@ -1804,10 +1804,6 @@ int tty_release(struct inode *inode, struct file *filp)
>>>  	 * each iteration we avoid any problems.
>>>  	 */
>>>  	while (1) {
>>> -		/* Guard against races with tty->count changes elsewhere and
>>> -		   opens on /dev/tty */
>>> -
>>> -		mutex_lock(&tty_mutex);
>>>  		tty_lock_pair(tty, o_tty);
>>>  		tty_closing = tty->count <= 1;
>>>  		o_tty_closing = o_tty &&
>>> @@ -1840,7 +1836,6 @@ int tty_release(struct inode *inode, struct file *filp)
>>>  		printk(KERN_WARNING "%s: %s: read/write wait queue active!\n",
>>>  				__func__, tty_name(tty, buf));
>>>  		tty_unlock_pair(tty, o_tty);
>>> -		mutex_unlock(&tty_mutex);
>>>  		schedule();
>>>  	}
>>>  
>>
>> The code in my tree in this section of tty_release() looks a bit
>> different, so I had to hand-apply this patch.
> 
> Although there's nothing wrong with your version, I'm wondering why this
> didn't apply cleanly.
> 
> While I go look at your tree, can you check that these patches are
> sitting on top of the earlier two patches you applied to your tty-linus
> branch; specifically 'tty: Fix high cpu load if tty is unreleasable' and
> 'tty: Prevent "read/write wait queue active!" log flooding'?

Yep, that's the problem: your 'tty-testing' branch doesn't have the 3 patches
from me that you put in your 'tty-linus' branch earlier this evening. Those are:

serial: Fix divide-by-zero fault in uart_get_divisor()
tty: Fix high cpu load if tty is unreleasable
tty: Prevent "read/write wait queue active!" log flooding

How can I help fix this?

Regards,
Peter Hurley


>> I've included the version
>> I used below, please verify I didn't mess it up.
>>
>> thanks,
>>
>> greg k-h
>>
>>
>>
>> diff --git a/drivers/tty/tty_io.c b/drivers/tty/tty_io.c
>> index ea8c6cae8d12..e59de81c39a9 100644
>> --- a/drivers/tty/tty_io.c
>> +++ b/drivers/tty/tty_io.c
>> @@ -1804,10 +1804,6 @@ int tty_release(struct inode *inode, struct file *filp)
>>  	 * each iteration we avoid any problems.
>>  	 */
>>  	while (1) {
>> -		/* Guard against races with tty->count changes elsewhere and
>> -		   opens on /dev/tty */
>> -
>> -		mutex_lock(&tty_mutex);
>>  		tty_lock_pair(tty, o_tty);
>>  		tty_closing = tty->count <= 1;
>>  		o_tty_closing = o_tty &&
>> @@ -1840,7 +1836,6 @@ int tty_release(struct inode *inode, struct file *filp)
>>  		printk(KERN_WARNING "%s: %s: read/write wait queue active!\n",
>>  				__func__, tty_name(tty, buf));
>>  		tty_unlock_pair(tty, o_tty);
>> -		mutex_unlock(&tty_mutex);
>>  		schedule();
>>  	}
>>  
>> @@ -1891,7 +1886,6 @@ int tty_release(struct inode *inode, struct file *filp)
>>  		read_unlock(&tasklist_lock);
>>  	}
>>  
>> -	mutex_unlock(&tty_mutex);
>>  	tty_unlock_pair(tty, o_tty);
>>  	/* At this point, the tty->count == 0 should ensure a dead tty
>>  	   cannot be re-opened by a racing opener */
>>
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-serial" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>