Re: Power domain clock warnings on recent next
From: Marek Szyprowski
Date: Wed Jan 17 2018 - 01:26:00 EST
Hi Krzysztof,
On 2018-01-16 21:07, Krzysztof Kozlowski wrote:
Hi everyone,
Anyone already noticed or started bisecting warnings coming from
clk_core_disable_lock/exynos_pd_power on recent linux-next?
Yes, I've noticed it early this week and prepared a patch:
https://www.spinics.net/lists/arm-kernel/msg628881.html
On next-20180116, Odroid XU3 and HC1 (Exynso5422):
[ 0.882736] EXYNOS5420 PMU initialized
[ 0.891979] ------------[ cut here ]------------
[ 0.892220] WARNING: CPU: 0 PID: 59 at ../drivers/clk/clk.c:811 clk_core_disable_lock+0x18/0x24
[ 0.892383] Modules linked in:
[ 0.892476] CPU: 0 PID: 59 Comm: kworker/0:1 Not tainted 4.15.0-rc8-next-20180116 #1121
[ 0.892629] Hardware name: SAMSUNG EXYNOS (Flattened Device Tree)
[ 0.892763] Workqueue: pm genpd_power_off_work_fn
[ 0.892891] [<c0111f90>] (unwind_backtrace) from [<c010e360>] (show_stack+0x10/0x14)
[ 0.893051] [<c010e360>] (show_stack) from [<c0a19860>] (dump_stack+0x98/0xc4)
[ 0.893209] [<c0a19860>] (dump_stack) from [<c012662c>] (__warn+0xf8/0x110)
[ 0.893353] [<c012662c>] (__warn) from [<c012675c>] (warn_slowpath_null+0x40/0x48)
[ 0.893511] [<c012675c>] (warn_slowpath_null) from [<c04b9a98>] (clk_core_disable_lock+0x18/0x24)
[ 0.893689] [<c04b9a98>] (clk_core_disable_lock) from [<c04ba820>] (clk_core_disable_unprepare+0xc/0x20)
[ 0.893874] [<c04ba820>] (clk_core_disable_unprepare) from [<c04ba87c>] (__clk_set_parent_after+0x48/0x4c)
[ 0.894063] [<c04ba87c>] (__clk_set_parent_after) from [<c04bb57c>] (clk_core_set_parent_nolock+0x27c/0x5e8)
[ 0.894253] [<c04bb57c>] (clk_core_set_parent_nolock) from [<c04bb920>] (clk_set_parent+0x38/0x6c)
[ 0.894438] [<c04bb920>] (clk_set_parent) from [<c04c9ea8>] (exynos_pd_power+0x88/0x1e4)
[ 0.894603] [<c04c9ea8>] (exynos_pd_power) from [<c0587ed8>] (genpd_power_off+0x160/0x274)
[ 0.894771] [<c0587ed8>] (genpd_power_off) from [<c0588a10>] (genpd_power_off_work_fn+0x2c/0x40)
[ 0.894953] [<c0588a10>] (genpd_power_off_work_fn) from [<c01465b4>] (process_one_work+0x2d4/0x8f0)
[ 0.895132] [<c01465b4>] (process_one_work) from [<c0147860>] (worker_thread+0x38/0x584)
[ 0.895297] [<c0147860>] (worker_thread) from [<c014d794>] (kthread+0x138/0x168)
[ 0.895450] [<c014d794>] (kthread) from [<c01010b4>] (ret_from_fork+0x14/0x20)
Later there is also a lockdep warning but it looks like second issue:
Right, this deplock issue is already known for over 2 months. It is a
false positive:
https://www.spinics.net/lists/linux-samsung-soc/msg61425.html
[ 5.932966] ======================================================
[ 5.937199] usb 5-1: new high-speed USB device number 2 using xhci-hcd
[ 5.939073] WARNING: possible circular locking dependency detected
[ 5.939110] 4.15.0-rc8-next-20180116 #1121 Tainted: G W
[ 5.958143] ------------------------------------------------------
[ 5.964299] kworker/0:1/59 is trying to acquire lock:
[ 5.969304] (&genpd->mlock){+.+.}, at: [<6abc3872>] genpd_runtime_resume+0x104/0x260
[ 5.977155]
[ 5.977155] but task is already holding lock:
[ 5.982926] (prepare_lock){+.+.}, at: [<74cef905>] clk_prepare_lock+0x10/0xf8
[ 5.990143]
[ 5.990143] which lock already depends on the new lock.
[ 5.990143]
[ 5.998309]
[ 5.998309] the existing dependency chain (in reverse order) is:
[ 6.005739]
[ 6.005739] -> #1 (prepare_lock){+.+.}:
[ 6.011042] mutex_lock_nested+0x1c/0x24
[ 6.015419] clk_prepare_lock+0x50/0xf8
[ 6.019755] clk_unprepare+0x1c/0x2c
[ 6.023841] exynos_pd_power+0x1a8/0x1e4
[ 6.028246] genpd_power_off+0x160/0x274
[ 6.032664] genpd_power_off_work_fn+0x2c/0x40
[ 6.037630] process_one_work+0x2d4/0x8f0
[ 6.042104] worker_thread+0x38/0x584
[ 6.046268] kthread+0x138/0x168
[ 6.049981] ret_from_fork+0x14/0x20
[ 6.054044] (null)
[ 6.056794]
[ 6.056794] -> #0 (&genpd->mlock){+.+.}:
[ 6.062238] __mutex_lock+0x7c/0xa68
[ 6.066278] mutex_lock_nested+0x1c/0x24
[ 6.070703] genpd_runtime_resume+0x104/0x260
[ 6.075557] __rpm_callback+0xc0/0x21c
[ 6.079792] rpm_callback+0x20/0x80
[ 6.083774] rpm_resume+0x558/0x7dc
[ 6.087762] __pm_runtime_resume+0x60/0x98
[ 6.092367] clk_core_prepare+0x44/0x490
[ 6.096783] clk_prepare+0x20/0x30
[ 6.100674] amba_get_enable_pclk+0x2c/0x60
[ 6.105363] amba_device_try_add+0x8c/0x20c
[ 6.110041] amba_deferred_retry_func+0x40/0xbc
[ 6.115080] process_one_work+0x2d4/0x8f0
[ 6.119569] worker_thread+0x38/0x584
[ 6.123727] kthread+0x138/0x168
[ 6.127444] ret_from_fork+0x14/0x20
[ 6.131510] (null)
[ 6.134263]
[ 6.134263] other info that might help us debug this:
[ 6.134263]
[ 6.142328] Possible unsafe locking scenario:
[ 6.142328]
[ 6.148178] CPU0 CPU1
[ 6.152656] ---- ----
[ 6.157160] lock(prepare_lock);
[ 6.160439] lock(&genpd->mlock);
[ 6.166365] lock(prepare_lock);
[ 6.172168] lock(&genpd->mlock);
[ 6.175517]
[ 6.175517] *** DEADLOCK ***
[ 6.175517]
[ 6.181475] 4 locks held by kworker/0:1/59:
[ 6.185580] #0: ((wq_completion)"events"){+.+.}, at: [<f71c19aa>] process_one_work+0x210/0x8f0
[ 6.194407] #1: ((deferred_retry_work).work){+.+.}, at: [<f71c19aa>] process_one_work+0x210/0x8f0
[ 6.203422] #2: (deferred_devices_lock){+.+.}, at: [<3e940c1f>] amba_deferred_retry_func+0x1c/0xbc
[ 6.212522] #3: (prepare_lock){+.+.}, at: [<74cef905>] clk_prepare_lock+0x10/0xf8
[ 6.220128]
[ 6.220128] stack backtrace:
[ 6.224438] CPU: 0 PID: 59 Comm: kworker/0:1 Tainted: G W 4.15.0-rc8-next-20180116 #1121
[ 6.233757] Hardware name: SAMSUNG EXYNOS (Flattened Device Tree)
[ 6.239791] Workqueue: events amba_deferred_retry_func
[ 6.244929] [<c0111f90>] (unwind_backtrace) from [<c010e360>] (show_stack+0x10/0x14)
[ 6.252670] [<c010e360>] (show_stack) from [<c0a19860>] (dump_stack+0x98/0xc4)
[ 6.259877] [<c0a19860>] (dump_stack) from [<c0181478>] (print_circular_bug.constprop.17+0x210/0x32c)
[ 6.269077] [<c0181478>] (print_circular_bug.constprop.17) from [<c01848f8>] (__lock_acquire+0x155c/0x1ac8)
[ 6.278786] [<c01848f8>] (__lock_acquire) from [<c0185884>] (lock_acquire+0xe0/0x2bc)
[ 6.286558] [<c0185884>] (lock_acquire) from [<c0a31788>] (__mutex_lock+0x7c/0xa68)
[ 6.294181] [<c0a31788>] (__mutex_lock) from [<c0a32190>] (mutex_lock_nested+0x1c/0x24)
[ 6.302161] [<c0a32190>] (mutex_lock_nested) from [<c05885dc>] (genpd_runtime_resume+0x104/0x260)
[ 6.311008] [<c05885dc>] (genpd_runtime_resume) from [<c057c6c4>] (__rpm_callback+0xc0/0x21c)
[ 6.319484] [<c057c6c4>] (__rpm_callback) from [<c057c840>] (rpm_callback+0x20/0x80)
[ 6.327185] [<c057c840>] (rpm_callback) from [<c057c2a8>] (rpm_resume+0x558/0x7dc)
[ 6.334721] [<c057c2a8>] (rpm_resume) from [<c057c58c>] (__pm_runtime_resume+0x60/0x98)
[ 6.342706] [<c057c58c>] (__pm_runtime_resume) from [<c04b7f6c>] (clk_core_prepare+0x44/0x490)
[ 6.351297] [<c04b7f6c>] (clk_core_prepare) from [<c04ba304>] (clk_prepare+0x20/0x30)
[ 6.359081] [<c04ba304>] (clk_prepare) from [<c04b52f4>] (amba_get_enable_pclk+0x2c/0x60)
[ 6.367229] [<c04b52f4>] (amba_get_enable_pclk) from [<c04b5510>] (amba_device_try_add+0x8c/0x20c)
[ 6.376164] [<c04b5510>] (amba_device_try_add) from [<c04b56f8>] (amba_deferred_retry_func+0x40/0xbc)
[ 6.385361] [<c04b56f8>] (amba_deferred_retry_func) from [<c01465b4>] (process_one_work+0x2d4/0x8f0)
[ 6.394457] [<c01465b4>] (process_one_work) from [<c0147860>] (worker_thread+0x38/0x584)
[ 6.402497] [<c0147860>] (worker_thread) from [<c014d794>] (kthread+0x138/0x168)
[ 6.409852] [<c014d794>] (kthread) from [<c01010b4>] (ret_from_fork+0x14/0x20)
Best regards
--
Marek Szyprowski, PhD
Samsung R&D Institute Poland