The serial core aquires the port spinlock before calling
port->ops->start_tx(), so sa1100_start_tx() shouldn't try to lock it
again.

BUG: spinlock recursion on CPU#0, init/1
 lock: c0205f20, .magic: dead4ead, .owner: init/1, .owner_cpu: 0
[<c0022cdc>] (dump_stack+0x0/0x14) [<c00dc338>] (spin_bug+0x0/0xbc)
[<c00dc6b0>] (_raw_spin_lock+0x0/0x170)  r8 = 00000007  r7 = C02FE0070
[<c018a2a8>] (_spin_lock_irqsave+0x0/0x24)  r4 = C0205F20
[<c0112110>] (sa1100_start_tx+0x0/0x40)  r4 = C038C000
[<c010ee38>] (__uart_start+0x0/0x5c) [<c010ee94>] (uart_start+0x0/0x3
[<c010f1d0>] (uart_write+0x0/0xdc) [<c00fee34>] (write_chan+0x0/0x370
Signed-off-by: Florin Malita <fmalita@gmail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>