Re: printk() vs tty_io

[Peter Zijlstra]

On Wed, 2011-12-14 at 09:05 -0500, Ted Ts'o wrote:
> On Wed, Dec 14, 2011 at 11:56:46AM +0100, Peter Zijlstra wrote:
> > 
> > Well, this was the 8250 serial port driver doing wakeups. Its a simple
> > on-board serial port,
> > 
> > [   29.797007]        [<ffffffff8108a650>] __wake_up_common+0x4e/0x84
> > [   29.797007]        [<ffffffff8108d6ea>] __wake_up+0x39/0x4d
> > [   29.797007]        [<ffffffff8135d0db>] tty_wakeup+0x5b/0x60
> > [   29.797007]        [<ffffffff81373fb9>] uart_write_wakeup+0x21/0x23
> > [   29.797007]        [<ffffffff81376899>] transmit_chars+0xd8/0x12f
> > [   29.797007]        [<ffffffff81377f3a>] serial8250_handle_port+0x2d9/0x2fd
> > [   29.797007]        [<ffffffff81377fcb>] serial8250_handle_irq+0x16/0x1d
> > [   29.797007]        [<ffffffff81377ff5>] serial8250_default_handle_irq+0x23/0x27
> > [   29.797007]        [<ffffffff81377ade>] serial8250_interrupt+0x4d/0xc6
> > [   29.797007]        [<ffffffff810c9ae8>] handle_irq_event_percpu+0xab/0x1fc
> > [   29.797007]        [<ffffffff810c9c7a>] handle_irq_event+0x41/0x61
> > [   29.797007]        [<ffffffff810cc4cd>] handle_edge_irq+0xd1/0xf6
> > [   29.797007]        [<ffffffff8103ab85>] handle_irq+0x24/0x2f
> > [   29.797007]        [<ffffffff8150acdd>] do_IRQ+0x4d/0xb3
> > 
> > which places the wakeup under uport->lock, serial8250_console_write()
> > takes uport->lock, under console_sem, connecting the locks.
> 
> printk() goes through serial8250_console_write() which writes to the
> UART device using a busy loop, with local IRQ's disabled.  So it's not
> serial8250_console_write() which is waking things up; this is coming
> form interrupt handler running on some different CPU.

Which can still deadlock you just fine..
--
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/