Re: [BUG] on sh7372 (mackerel) with today's Linus' (github) master

["Rafael J. Wysocki" <rjw@sisk.pl>]

Hi,

On Friday, September 23, 2011, Guennadi Liakhovetski wrote:
> Hi
> 
> Not a very pleasant message from an -rc7 kernel:

Well, I wonder why I can't reproduce this.

> [    9.906250] BUG: sleeping function called from invalid context at /home/lyakh/software/project/24/src/linux-2.6/kernel/mutex.c:271
> [    9.906250] in_atomic(): 1, irqs_disabled(): 128, pid: 167, name: modprobe
> [    9.906250] 3 locks held by modprobe/167:
> [    9.906250]  #0:  (&__lockdep_no_validate__){......}, at: [<c0176114>] __driver_attach+0x48/0x94
> [    9.906250]  #1:  (&__lockdep_no_validate__){......}, at: [<c0176124>] __driver_attach+0x58/0x94
> [    9.906250]  #2:  (&(&pcd->lock)->rlock){......}, at: [<c017f0d4>] pm_clk_resume+0x28/0x88
> [    9.906250] Backtrace: 
> [    9.906250] [<c0012434>] (dump_backtrace+0x0/0x110) from [<c024f8f0>] (dump_stack+0x18/0x1c)
> [    9.906250]  r6:00000000 r5:cf0dde20 r4:cf798000 r3:60000093
> [    9.906250] [<c024f8d8>] (dump_stack+0x0/0x1c) from [<c001da88>] (__might_sleep+0x100/0x120)
> [    9.906250] [<c001d988>] (__might_sleep+0x0/0x120) from [<c025192c>] (mutex_lock_nested+0x2c/0x2fc)
> [    9.906250]  r4:c0356808
> [    9.906250] [<c0251900>] (mutex_lock_nested+0x0/0x2fc) from [<c01a5a44>] (clk_get_sys+0x2c/0xf4)
> [    9.906250]  r8:00000000 r7:00000000 r6:00000000 r5:cf0dde20 r4:c0356808
> [    9.906250] [<c01a5a18>] (clk_get_sys+0x0/0xf4) from [<c01a5b34>] (clk_get+0x28/0x2c)
> [    9.906250] [<c01a5b0c>] (clk_get+0x0/0x2c) from [<c017f094>] (pm_clk_acquire+0x18/0x30)
> [    9.906250] [<c017f07c>] (pm_clk_acquire+0x0/0x30) from [<c017f0f8>] (pm_clk_resume+0x4c/0x88)
> [    9.906250]  r4:cf0dcf40 r3:00000000
> [    9.906250] [<c017f0ac>] (pm_clk_resume+0x0/0x88) from [<c017c6b8>] (rpm_callback+0x4c/0x6c)
> [    9.906250]  r8:cf798000 r7:c0333be0 r6:c0333be0 r5:c0333b50 r4:c017f0ac
> [    9.906250] r3:00000001
> [    9.906250] [<c017c66c>] (rpm_callback+0x0/0x6c) from [<c017d3f0>] (rpm_resume+0x300/0x3d4)
> [    9.906250]  r6:00000000 r5:c03520c8 r4:c0333b50 r3:c0331fb8
> [    9.906250] [<c017d0f0>] (rpm_resume+0x0/0x3d4) from [<c017d750>] (__pm_runtime_resume+0x50/0x68)
> [    9.906250] [<c017d700>] (__pm_runtime_resume+0x0/0x68) from [<bf016834>] (sh_mobile_i2c_xfer+0x3c/0x37c [i2c_sh_mobile])
> [    9.906250]  r7:00000002 r6:00000009 r5:cf799cc0 r4:cf64f800
> [    9.906250] [<bf0167f8>] (sh_mobile_i2c_xfer+0x0/0x37c [i2c_sh_mobile]) from [<bf0016ec>] (i2c_transfer+0x98/0xe4 [i2c_core])
> [    9.906250] [<bf001654>] (i2c_transfer+0x0/0xe4 [i2c_core]) from [<bf001bb0>] (i2c_smbus_xfer+0x3cc/0x504 [i2c_core])
> [    9.906250] [<bf0017e4>] (i2c_smbus_xfer+0x0/0x504 [i2c_core]) from [<bf001fdc>] (i2c_smbus_read_byte_data+0x3c/0x4c [i2c_core])
> [    9.906250] [<bf001fa0>] (i2c_smbus_read_byte_data+0x0/0x4c [i2c_core]) from [<bf0910d8>] (tca6416_read_reg+0x48/0x8c [tca6416_keypad])
> [    9.906250] [<bf091090>] (tca6416_read_reg+0x0/0x8c [tca6416_keypad]) from [<bf091500>] (tca6416_keypad_probe+0x1dc/0x3a0 [tca6416_keypad])
> [    9.906250]  r7:00000030 r6:c033d204 r5:cf3dd800 r4:cf1ff0a0
> [    9.906250] [<bf091324>] (tca6416_keypad_probe+0x0/0x3a0 [tca6416_keypad]) from [<bf0005cc>] (i2c_device_probe+0x7c/0xa4 [i2c_core])
> [    9.906250] [<bf000550>] (i2c_device_probe+0x0/0xa4 [i2c_core]) from [<c0176010>] (driver_probe_device+0xd0/0x18c)
> [    9.906250]  r6:bf091878 r5:cf64f054 r4:cf64f020 r3:bf000550
> [    9.906250] [<c0175f40>] (driver_probe_device+0x0/0x18c) from [<c017613c>] (__driver_attach+0x70/0x94)
> [    9.906250]  r7:00000000 r6:bf091878 r5:cf64f054 r4:cf64f020
> [    9.906250] [<c01760cc>] (__driver_attach+0x0/0x94) from [<c01757a4>] (bus_for_each_dev+0x54/0x84)
> [    9.906250]  r6:00000000 r5:c01760cc r4:bf091878 r3:00000000
> [    9.906250] [<c0175750>] (bus_for_each_dev+0x0/0x84) from [<c0175e34>] (driver_attach+0x20/0x28)
> [    9.906250]  r6:bf002df8 r5:cf3ef440 r4:bf091878
> [    9.906250] [<c0175e14>] (driver_attach+0x0/0x28) from [<c017506c>] (bus_add_driver+0xa8/0x228)
> [    9.906250] [<c0174fc4>] (bus_add_driver+0x0/0x228) from [<c01767c8>] (driver_register+0xb0/0x140)
> [    9.906250] [<c0176718>] (driver_register+0x0/0x140) from [<bf0009a4>] (i2c_register_driver+0x48/0xb4 [i2c_core])
> [    9.906250]  r8:00000001 r7:00000000 r6:00000001 r5:bf093000 r4:bf091850
> [    9.906250] r3:bf002df8
> [    9.906250] [<bf00095c>] (i2c_register_driver+0x0/0xb4 [i2c_core]) from [<bf093018>] (tca6416_keypad_init+0x18/0x24 [tca6416_keypad])
> [    9.906250]  r5:bf093000 r4:bf0918c4
> [    9.906250] [<bf093000>] (tca6416_keypad_init+0x0/0x24 [tca6416_keypad]) from [<c0009374>] (do_one_initcall+0xa0/0x170)
> [    9.906250] [<c00092d4>] (do_one_initcall+0x0/0x170) from [<c005583c>] (sys_init_module+0xd24/0xf48)
> [    9.906250]  r9:0000001b r8:00000001 r7:cf5793a0 r6:00000001 r5:bf09190c
> [    9.906250] r4:bf0918c4
> [    9.906250] [<c0054b18>] (sys_init_module+0x0/0xf48) from [<c000efc0>] (ret_fast_syscall+0x0/0x30)
> [    9.960937] input: tca6408-keys as /devices/platform/i2c-sh_mobile.0/i2c-0/0-0020/input/input0

There is a bug in pm_clk_suspend() (and analogously in pm_clk_resume())
that calls pm_clk_acquire() under a spinlock, but pm_clk_acquire()
calls clk_get().

I'm not sure how to fix this at the moment.  We may need to revert commits
b7ab83e and 5a50a01, although I would hate that, because it would require us
to fix sh-sci runtime PM in a different way.

I'll try to figure out something over the weekend.

Thanks,
Rafael