On Wed, 2007-07-25 at 19:00 +0200, Juergen Beisert wrote: > [c0245db0] [c01bdb98] rt_spin_lock_slowlock+0x4c/0x224 (unreliable) > [c0245e10] [c011823c] uart_start+0x24/0x48 > [c0245e30] [c0113ff4] n_tty_receive_buf+0x170/0xfd4 > [c0245ef0] [c010f0dc] flush_to_ldisc+0xe0/0x130 > [c0245f20] [c011b51c] mpc52xx_uart_int+0x194/0x350 > [c0245f50] [c0046dfc] handle_IRQ_event+0x6c/0x110 > [c0245f80] [c00475ec] thread_simple_irq+0x90/0xf8 > [c0245fa0] [c00479a0] do_irqd+0x34c/0x3cc > [c0245fd0] [c0033380] kthread+0x48/0x84 > [c0245ff0] [c00104ac] kernel_thread+0x44/0x60 > Instruction dump: > 70090008 40820144 80010064 bb410048 38210060 7c0803a6 4e800020 801c0010 > 5400003a 7c001278 7c000034 5400d97e <0f000000> 39600004 91610008 80010008 > note: IRQ-131[93] exited with preempt_count 1 Yup. That's a deadlock. In mainline this does not happen, as the spinlock is a NOP. Turn on CONFIG_PROVE_LOCKING in mainline and you see the problem as well. Solution below tglx Index: linux-2.6.22/drivers/serial/mpc52xx_uart.c =================================================================== --- linux-2.6.22.orig/drivers/serial/mpc52xx_uart.c 2007-07-09 01:32:17.000000000 +0200 +++ linux-2.6.22/drivers/serial/mpc52xx_uart.c 2007-07-25 21:06:11.000000000 +0200 @@ -501,7 +501,9 @@ mpc52xx_uart_int_rx_chars(struct uart_po } } + spin_unlock(&port->lock); tty_flip_buffer_push(tty); + spin_lock(&port->lock); return in_be16(&PSC(port)->mpc52xx_psc_status) & MPC52xx_PSC_SR_RXRDY; } - To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
