[ Upstream commit 063773011d33bb36588a90385aa9eb75d13c6d80 ]
Lockdep reports the following issue on my setup:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock((work_completion)(&(&rdev->disable_work)->work));
lock(regulator_list_mutex);
lock((work_completion)(&(&rdev->disable_work)->work));
lock(regulator_list_mutex);
The problem is that regulator_unregister takes the
regulator_list_mutex and then calls flush_work on disable_work. But
regulator_disable_work calls regulator_lock_dependent which will
also take the regulator_list_mutex. Resulting in a deadlock if the
flush_work call actually needs to flush the work.
Fix this issue by moving the flush_work outside of the
regulator_list_mutex. The list mutex is not used to guard the point at
which the delayed work is queued, so its use adds no additional safety.
Fixes: f8702f9e4aa7 ("regulator: core: Use ww_mutex for regulators locking")
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 70b464918e5331e488058870fcc6821d54c4e541 upstream.
During several error paths in the function
regulator_set_voltage_unlocked() the value of 'ret' can take on negative
error values. However, in calls that go through the 'goto out' statement,
this return value is lost and return 0 is used instead, indicating a
'pass'.
There are several cases where this function should legitimately return a
fail instead of a pass: one such case includes constraints check during
voltage selection in the call to regulator_check_voltage(), which can
have -EINVAL for the case when an unsupported voltage is incorrectly
requested. In that case, -22 is expected as the return value, not 0.
Fixes: 9243a195be7a ("regulator: core: Change voltage setting path")
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Steve Twiss <stwiss.opensource@diasemi.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit e5e21f70bfd3a201e627b48aed82793d1bcd6f78 ]
Take the regulator lock before applying system load.
Fixes the following lockdep splat:
[ 5.583581] WARNING: CPU: 1 PID: 16 at drivers/regulator/core.c:925 drms_uA_update+0x114/0x360
[ 5.588467] Modules linked in:
[ 5.596833] CPU: 1 PID: 16 Comm: kworker/1:0 Not tainted 5.0.0-rc6-next-20190213-00002-g0fce66ab480f #18
[ 5.599933] Hardware name: Qualcomm Technologies, Inc. APQ 8016 SBC (DT)
[ 5.609544] Workqueue: events qcom_channel_state_worker
[ 5.616209] pstate: 60000005 (nZCv daif -PAN -UAO)
[ 5.621152] pc : drms_uA_update+0x114/0x360
[ 5.626006] lr : drms_uA_update+0x110/0x360
[ 5.630084] sp : ffff0000124b3490
[ 5.634242] x29: ffff0000124b3490 x28: ffff800005326e00
[ 5.637735] x27: ffff0000124b35f8 x26: 000000000032bc48
[ 5.643117] x25: ffff800004c7e800 x24: ffff800004c6d500
[ 5.648411] x23: ffff800004c38a80 x22: 00000000000000d1
[ 5.653706] x21: 00000000001ab3f0 x20: ffff800004c7e800
[ 5.659001] x19: ffff0000114c3000 x18: ffffffffffffffff
[ 5.664297] x17: 0000000000000000 x16: 0000000000000000
[ 5.669592] x15: ffff0000114c3808 x14: 0720072007200720
[ 5.674888] x13: 00000000199c9b28 x12: ffff80002bcccc40
[ 5.680183] x11: ffff000012286000 x10: ffff0000114c3808
[ 5.685477] x9 : 0720072007200720 x8 : ffff000010e9e808
[ 5.690772] x7 : ffff0000106da568 x6 : 0000000000000000
[ 5.696067] x5 : 0000000000000000 x4 : 0000000000000000
[ 5.701362] x3 : 0000000000000004 x2 : 0000000000000000
[ 5.706658] x1 : 0000000000000000 x0 : 0000000000000000
[ 5.711952] Call trace:
[ 5.717223] drms_uA_update+0x114/0x360
[ 5.719405] regulator_register+0xb30/0x1140
[ 5.723230] devm_regulator_register+0x4c/0xa8
[ 5.727745] rpm_reg_probe+0xfc/0x1b0
[ 5.731992] platform_drv_probe+0x50/0xa0
[ 5.735727] really_probe+0x20c/0x2b8
[ 5.739718] driver_probe_device+0x58/0x100
[ 5.743368] __device_attach_driver+0x90/0xd0
[ 5.747363] bus_for_each_drv+0x64/0xc8
[ 5.751870] __device_attach+0xd8/0x138
[ 5.755516] device_initial_probe+0x10/0x18
[ 5.759341] bus_probe_device+0x98/0xa0
[ 5.763502] device_add+0x3d0/0x640
[ 5.767319] of_device_add+0x48/0x58
[ 5.770793] of_platform_device_create_pdata+0xb0/0x128
[ 5.774629] of_platform_bus_create+0x174/0x370
[ 5.779569] of_platform_populate+0x78/0xe0
[ 5.784082] qcom_smd_rpm_probe+0x80/0xa0
[ 5.788245] rpmsg_dev_probe+0x114/0x1a0
[ 5.792411] really_probe+0x20c/0x2b8
[ 5.796401] driver_probe_device+0x58/0x100
[ 5.799964] __device_attach_driver+0x90/0xd0
[ 5.803960] bus_for_each_drv+0x64/0xc8
[ 5.808468] __device_attach+0xd8/0x138
[ 5.812115] device_initial_probe+0x10/0x18
[ 5.815936] bus_probe_device+0x98/0xa0
[ 5.820099] device_add+0x3d0/0x640
[ 5.823916] device_register+0x1c/0x28
[ 5.827391] rpmsg_register_device+0x4c/0x90
[ 5.831216] qcom_channel_state_worker+0x170/0x298
[ 5.835651] process_one_work+0x294/0x6e8
[ 5.840241] worker_thread+0x40/0x450
[ 5.844318] kthread+0x11c/0x120
[ 5.847961] ret_from_fork+0x10/0x18
[ 5.851260] irq event stamp: 9090
[ 5.854820] hardirqs last enabled at (9089): [<ffff000010160798>] console_unlock+0x3e0/0x5b0
[ 5.858086] hardirqs last disabled at (9090): [<ffff0000100817cc>] do_debug_exception+0x104/0x140
[ 5.866596] softirqs last enabled at (9086): [<ffff000010082024>] __do_softirq+0x474/0x574
[ 5.875446] softirqs last disabled at (9079): [<ffff0000100f2254>] irq_exit+0x13c/0x148
[ 5.883598] ---[ end trace 6984ef7f081afa21 ]---
Fixes: fa94e48e13a1 ("regulator: core: Apply system load even if no consumer loads")
Signed-off-by: Niklas Cassel <niklas.cassel@linaro.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Use DEFINE_SHOW_ATTRIBUTE macro to simplify the code. Signed-off-by: Yangtao Li <tiny.windzz@gmail.com> Signed-off-by: Mark Brown <broonie@kernel.org>
At the end of regulator_resolve_supply() we have historically turned
on our supply in some cases. This could be for one of two reasons:
1. If resolving supplies was happening before the call to
set_machine_constraints() we needed to predict if
set_machine_constraints() was going to turn the regulator on and we
needed to preemptively turn the supply on.
2. Maybe set_machine_constraints() happened before we could resolve
supplies (because we failed the first time to resolve) and thus we
might need to propagate an enable that already happened up to our
supply.
Historically regulator_resolve_supply() used _regulator_is_enabled()
to decide whether to turn on the supply.
Let's change things a little bit. Specifically:
1. Let's try to enable the supply and the regulator in the same place,
both in set_machine_constraints(). This means that we have exactly
the same logic for enabling the supply and the regulator.
2. Let's properly set use_count when we enable always-on or boot-on
regulators even for those that don't have supplies. The previous
commit 1fc12b05895e ("regulator: core: Avoid propagating to
supplies when possible") only did this right for regulators with
supplies.
3. Let's make it clear that the only time we need to enable the supply
in regulator_resolve_supply() is if the main regulator is currently
in use. By using use_count (like the rest of the code) to decide
if we're going to enable our supply we keep everything consistent.
Overall the new scheme should be cleaner and easier to reason about.
In addition to fixing regulator_summary to be more correct (because of
the more correct use_count), this change also has the effect of no
longer using _regulator_is_enabled() in this code path.
_regulator_is_enabled() could return an error code for some regulators
at bootup (like RPMh) that can't read their initial state. While one
can argue that the design of those regulators is sub-optimal, the new
logic sidesteps this brokenness. This fix in particular fixes
observed problems on Qualcomm sdm845 boards which use the
above-mentioned RPMh regulator. Those problems were made worse by
commit 1fc12b05895e ("regulator: core: Avoid propagating to supplies
when possible") because now we'd think at bootup that the SD
regulators were already enabled and we'd never try them again.
Fixes: 1fc12b05895e ("regulator: core: Avoid propagating to supplies when possible")
Reported-by: Evan Green <evgreen@chromium.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
If a GPIO descriptor is passed to the regulator_register() function inside the config->ena_gpiod callers must be sure that once they call this API the regulator core owns that descriptor and will make sure to issue gpiod_put() on it, no matter whether the call is successful or not. For device tree regulators, the regulator core will automatically set up regulator init data from the device tree when registering a regulator by calling regulator_of_get_init_data() which in turn calls down to the regulator driver's .of_parse_cb() callback. This callback (in drivers such as for max77686) may also choose to fill in the config->ena_gpiod field with a GPIO descriptor. Harden the errorpath of regulator_register() to properly gpiod_put() any passed in cfg->ena_gpiod or any gpiod coming from the device tree on any type of error. Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Reviewed-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Reviewed-by: Charles Keepax <ckeepax@opensource.cirrus.com> Signed-off-by: Mark Brown <broonie@kernel.org>
Prior to commit 5451781dadf8 ("regulator: core: Only count load for
enabled consumers") we used to always add up the total load on every
enable in _regulator_enable(). After that commit we only updated the
total load when enabling / disabling a regulator where a consumer
specified a load or when changing the consumer load on an enabled
regulator.
The problem with the new scheme is that if there is a system load
specified for a regulator but no consumers specify a load then we
never account for it.
Let's account for the system load in set_machine_constraints().
NOTE: with the new scheme we end up with a bit of a quandry. What if
someone specifies _both_ an initial mode and a system load? If we
take the system load into account right at init time then it will
effectively clobber the initial mode. We'll resolve this by saying
that if both are specified then the initial mode will win. The system
load will then only take effect if/when a consumer specifies a load.
If no consumers ever specify a load then the initial mode will persist
and the system load will have no effect.
Fixes: 5451781dadf8 ("regulator: core: Only count load for enabled consumers")
Reported-by: Brian Masney <masneyb@onstation.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Brian Masney <masneyb@onstation.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
When a regulator is marked as always on, it is enabled early on, when checking and setting up constraints. It makes the assumption that the bootloader properly initialized the regulator, and just in case enables the regulator anyway. Some constraints however currently get missed, such as the soft-start and ramp-delay. This causes the regulator to be enabled, without the soft-start and ramp-delay being applied, which in turn can cause high-currents or other start-up problems. By moving the always-enabled constraints later in the constraints check, we can at least ensure all constraints for the regulator are followed. Signed-off-by: Olliver Schinagl <oliver@schinagl.nl> Signed-off-by: Priit Laes <plaes@plaes.org> Signed-off-by: Mark Brown <broonie@kernel.org>