The current driver triggers a lockdep warning for if tty_flip_buffer_push() is
called with uart_port->lock locked. This never shows up on UP kernels and comes
up only on SMP kernels.
Crash looks like this (produced with samsung.c driver):
-----
[<
c0014d58>] (unwind_backtrace+0x0/0xf8) from [<
c0011908>] (show_stack+0x10/0x14)
[<
c0011908>] (show_stack+0x10/0x14) from [<
c035da34>] (dump_stack+0x6c/0xac)
[<
c035da34>] (dump_stack+0x6c/0xac) from [<
c01b59ac>] (do_raw_spin_unlock+0xc4/0xd8)
[<
c01b59ac>] (do_raw_spin_unlock+0xc4/0xd8) from [<
c03627e4>] (_raw_spin_unlock_irqrestore+0xc/0)
[<
c03627e4>] (_raw_spin_unlock_irqrestore+0xc/0x38) from [<
c020a1a8>] (s3c24xx_serial_rx_chars+0)
[<
c020a1a8>] (s3c24xx_serial_rx_chars+0x12c/0x260) from [<
c020aae8>] (s3c64xx_serial_handle_irq+)
[<
c020aae8>] (s3c64xx_serial_handle_irq+0x48/0x60) from [<
c006aaa0>] (handle_irq_event_percpu+0x)
[<
c006aaa0>] (handle_irq_event_percpu+0x50/0x194) from [<
c006ac20>] (handle_irq_event+0x3c/0x5c)
[<
c006ac20>] (handle_irq_event+0x3c/0x5c) from [<
c006d864>] (handle_fasteoi_irq+0x80/0x13c)
[<
c006d864>] (handle_fasteoi_irq+0x80/0x13c) from [<
c006a4a4>] (generic_handle_irq+0x20/0x30)
[<
c006a4a4>] (generic_handle_irq+0x20/0x30) from [<
c000f454>] (handle_IRQ+0x38/0x94)
[<
c000f454>] (handle_IRQ+0x38/0x94) from [<
c0008538>] (gic_handle_irq+0x34/0x68)
[<
c0008538>] (gic_handle_irq+0x34/0x68) from [<
c00123c0>] (__irq_svc+0x40/0x70)
Exception stack(0xc04cdf70 to 0xc04cdfb8)
df60:
00000000 00000000 0000166e 00000000
df80:
c04cc000 c050278f c050278f 00000001 c04d444c 410fc0f4 c03649b0 00000000
dfa0:
00000001 c04cdfb8 c000f758 c000f75c 60070013 ffffffff
[<
c00123c0>] (__irq_svc+0x40/0x70) from [<
c000f75c>] (arch_cpu_idle+0x28/0x30)
[<
c000f75c>] (arch_cpu_idle+0x28/0x30) from [<
c0054888>] (cpu_startup_entry+0x5c/0x148)
[<
c0054888>] (cpu_startup_entry+0x5c/0x148) from [<
c0497aa4>] (start_kernel+0x334/0x38c)
BUG: spinlock lockup suspected on CPU#0, kworker/0:1/360
lock: s3c24xx_serial_ports+0x1d8/0x370, .magic:
dead4ead, .owner: <none>/-1, .owner_cpu: -1
CPU: 0 PID: 360 Comm: kworker/0:1 Not tainted 3.11.0-rc6-next-
20130819-00003-g75485f1 #2
Workqueue: events flush_to_ldisc
[<
c0014d58>] (unwind_backtrace+0x0/0xf8) from [<
c0011908>] (show_stack+0x10/0x14)
[<
c0011908>] (show_stack+0x10/0x14) from [<
c035da34>] (dump_stack+0x6c/0xac)
[<
c035da34>] (dump_stack+0x6c/0xac) from [<
c01b581c>] (do_raw_spin_lock+0x100/0x17c)
[<
c01b581c>] (do_raw_spin_lock+0x100/0x17c) from [<
c03628a0>] (_raw_spin_lock_irqsave+0x20/0x28)
[<
c03628a0>] (_raw_spin_lock_irqsave+0x20/0x28) from [<
c0203224>] (uart_start+0x18/0x34)
[<
c0203224>] (uart_start+0x18/0x34) from [<
c01ef890>] (__receive_buf+0x4b4/0x738)
[<
c01ef890>] (__receive_buf+0x4b4/0x738) from [<
c01efb44>] (n_tty_receive_buf2+0x30/0x98)
[<
c01efb44>] (n_tty_receive_buf2+0x30/0x98) from [<
c01f2ba8>] (flush_to_ldisc+0xec/0x138)
[<
c01f2ba8>] (flush_to_ldisc+0xec/0x138) from [<
c0031af0>] (process_one_work+0xfc/0x348)
[<
c0031af0>] (process_one_work+0xfc/0x348) from [<
c0032138>] (worker_thread+0x138/0x37c)
[<
c0032138>] (worker_thread+0x138/0x37c) from [<
c0037a7c>] (kthread+0xa4/0xb0)
[<
c0037a7c>] (kthread+0xa4/0xb0) from [<
c000e5f8>] (ret_from_fork+0x14/0x3c)
-----
Release the port lock before calling tty_flip_buffer_push() and reacquire it
after the call.
Similar stuff was already done for few other drivers in the past, like:
commit
2389b272168ceec056ca1d8a870a97fa9c26e11a
Author: Thomas Gleixner <tglx@linutronix.de>
Date: Tue May 29 21:53:50 2007 +0100
[ARM] 4417/1: Serial: Fix AMBA drivers locking
Cc: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>