bug in PL011 console

[Uwe Kleine-König]

Hello,

since mxs switched to amba-pl011 I often hit warnings like this:

	[    9.280000] udevd (74): /proc/74/oom_adj is deprecated, please use /proc/74/oom_score_adj instead.
	[    9.280000] BUG: sleeping function called from invalid context at /ptx/work/octopus/WORK_2_A/ukl/backup/gsrc/linux-2.6/kernel/mutex.c:278
	[    9.280000] in_atomic(): 0, irqs_disabled(): 128, pid: 74, name: udevd
	[    9.280000] 2 locks held by udevd/74:
	[    9.280000]  #0:  (&(&p->alloc_lock)->rlock){+.+...}, at: [<c0132ff8>] oom_adjust_write+0xf8/0x290
	[    9.280000]  #1:  (&(&sighand->siglock)->rlock){......}, at: [<c005d1ec>] __lock_task_sighand+0x64/0xac
	[    9.280000] irq event stamp: 673
	[    9.280000] hardirqs last  enabled at (672): [<c0033fdc>] vector_swi+0x3c/0x90
	[    9.280000] hardirqs last disabled at (673): [<c02882e0>] _raw_spin_lock_irqsave+0x30/0x64
	[    9.280000] softirqs last  enabled at (0): [<c004a000>] copy_process+0x3a8/0xf90
	[    9.280000] softirqs last disabled at (0): [<  (null)>]   (null)
	[    9.280000] Backtrace: 
	[    9.280000] [<c0037b54>] (dump_backtrace+0x0/0x110) from [<c02853a4>] (dump_stack+0x1c/0x20)
	[    9.280000]  r7:c79f0000 r6:c003dd00 r5:c030ce72 r4:c79f0000
	[    9.280000] [<c0285388>] (dump_stack+0x0/0x20) from [<c0046a58>] (__might_sleep+0x100/0x120)
	[    9.280000] [<c0046958>] (__might_sleep+0x0/0x120) from [<c0287054>] (mutex_lock_nested+0x3c/0x2b4)
	[    9.280000]  r5:00000000 r4:00000000
	[    9.280000] [<c0287018>] (mutex_lock_nested+0x0/0x2b4) from [<c003dd00>] (clk_enable+0x30/0x58)
	[    9.280000] [<c003dcd0>] (clk_enable+0x0/0x58) from [<c01d9888>] (pl011_console_write+0x30/0x8c)
	[    9.280000]  r4:00000066
	[    9.280000] [<c01d9858>] (pl011_console_write+0x0/0x8c) from [<c004b9a4>] (__call_console_drivers+0x68/0x84)
	[    9.280000]  r6:00000066 r5:00007401 r4:c0372950
	[    9.280000] [<c004b93c>] (__call_console_drivers+0x0/0x84) from [<c004ba48>] (_call_console_drivers+0x88/0x9c)
	[    9.280000]  r8:ffff8b99 r7:60000093 r6:c0352c24 r5:c0352ba0 r4:00007467
	[    9.280000] [<c004b9c0>] (_call_console_drivers+0x0/0x9c) from [<c004c040>] (release_console_sem+0x158/0x244)
	[    9.280000]  r5:00007467 r4:00007467
	[    9.280000] [<c004bee8>] (release_console_sem+0x0/0x244) from [<c004c7fc>] (vprintk+0x388/0x404)
	[    9.280000] [<c004c474>] (vprintk+0x0/0x404) from [<c0285560>] (printk+0x20/0x28)
	[    9.280000] [<c0285540>] (printk+0x0/0x28) from [<c01330f0>] (oom_adjust_write+0x1f0/0x290)
	[    9.280000]  r3:0000004a r2:0000004a r1:c795595c r0:c031998c
	[    9.280000] [<c0132f00>] (oom_adjust_write+0x0/0x290) from [<c00eb008>] (vfs_write+0xb8/0x18c)
	[    9.280000]  r8:0001a4ac r7:c79f1f70 r6:0001a4ac r5:00000003 r4:c79dbc00
	[    9.280000] [<c00eaf50>] (vfs_write+0x0/0x18c) from [<c00eb1a8>] (sys_write+0x48/0x74)
	[    9.280000]  r8:0001a4ac r7:00000003 r6:c79dbc00 r5:00000000 r4:00000000
	[    9.280000] [<c00eb160>] (sys_write+0x0/0x74) from [<c0033e80>] (ret_fast_syscall+0x0/0x38)
	[    9.280000]  r8:c0034088 r7:00000004 r6:00000000 r5:00000003 r4:00000003

Steven told me on irc that sleeping was not allowed in the console write
callback.  Maybe this didn't show up earlier because not all clk
implementations sleep as mxs' does.

I think the only possible fix is to do the clk_enable in the setup
callback instead of per-write.

Will send a patch as follow up.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |