gpio: max732x: Fix possible deadlock

This patch was derived from next one:
"gpio: fix pca953x set_type 'scheduling while atomic' bug".

After adding entry that consumes max732x GPIO as interrupt line to dts
file, deadlock appears somewhere in max732x probe function.

Deadlock caught by lockdep (from kernel log):
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< cut here >>>>>>>>>>>>>>>>>>>>>>>>>>>
[    0.473419] ======================================================
[    0.473419] [ INFO: HARDIRQ-safe -> HARDIRQ-unsafe lock order detected ]
[    0.473449] 3.x.xx-xxxxx-xxxxxxxx-dirty #2 Tainted: G        W
[    0.473449] ------------------------------------------------------
[    0.473449] swapper/0/1 [HC0[0]:SC0[0]:HE0:SE1] is trying to acquire:
[    0.473449]  (&lock->wait_lock){+.+...}, at: [<c072e350>] rt_mutex_trylock+0xc/0x74
[    0.473480]
[    0.473480] and this task is already holding:
[    0.473510]  (&chip->lock){......}, at: [<c0314514>] max732x_gpio_set_value+0x2c/0xa4
[    0.473541] which would create a new lock dependency:
[    0.473541]  (&chip->lock){......} -> (&lock->wait_lock){+.+...}

...

[    0.474273]  *** DEADLOCK ***
[    0.474273]
[    0.474273] 5 locks held by swapper/0/1:
[    0.474273]  #0:  (&__lockdep_no_validate__){......}, at: [<c03b2328>] __driver_attach+0x48/0x98
[    0.474304]  #1:  (&__lockdep_no_validate__){......}, at: [<c03b2338>] __driver_attach+0x58/0x98
[    0.474334]  #2:  (&chip->irq_lock){+.+...}, at: [<c0313e3c>] max732x_irq_bus_lock+0x14/0x20
[    0.474365]  #3:  (&irq_desc_lock_class){-.....}, at: [<c00a65a4>] __irq_get_desc_lock+0x48/0x88
[    0.474365]  #4:  (&chip->lock){......}, at: [<c0314514>] max732x_gpio
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< cut here >>>>>>>>>>>>>>>>>>>>>>>>>>>

Signed-off-by: Semen Protsenko <semen.protsenko@globallogic.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
This commit is contained in:
Semen Protsenko 2015-01-13 15:41:44 +02:00 • committed by Linus Walleij
commit 09afa276d5