[PATCH] regulator: core: Fix enable_count imbalance with EXCLUSIVE_GET

From: Zev Weiss
Date: Thu May 05 2022 - 00:32:34 EST


Since the introduction of regulator->enable_count, a driver that did
an exclusive get on an already-enabled regulator would end up with
enable_count initialized to 0 but rdev->use_count initialized to 1.
With that starting point the regulator is effectively stuck enabled,
because if the driver attempted to disable it it would fail the
enable_count underflow check in _regulator_handle_consumer_disable().

The EXCLUSIVE_GET path in _regulator_get() now initializes
enable_count along with rdev->use_count so that the regulator can be
disabled without underflowing the former.

Signed-off-by: Zev Weiss <zev@xxxxxxxxxxxxxxxxx>
Fixes: 5451781dadf85 ("regulator: core: Only count load for enabled consumers")
---

I tripped on this problem a bit in some recent experimentation with
the regulator API -- at first I figured I was misusing something, but
after looking a bit closer I *think* it's a bug (perhaps unnoticed
until now because EXCLUSIVE_GET doesn't seem to be very widely used),
but I'm not real knowledgeable about the regulator subsystem so
apologies if I've misdiagnosed.

drivers/regulator/core.c | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/regulator/core.c b/drivers/regulator/core.c
index d2553970a67b..c4d844ffad7a 100644
--- a/drivers/regulator/core.c
+++ b/drivers/regulator/core.c
@@ -2133,10 +2133,13 @@ struct regulator *_regulator_get(struct device *dev, const char *id,
rdev->exclusive = 1;

ret = _regulator_is_enabled(rdev);
- if (ret > 0)
+ if (ret > 0) {
rdev->use_count = 1;
- else
+ regulator->enable_count = 1;
+ } else {
rdev->use_count = 0;
+ regulator->enable_count = 0;
+ }
}

link = device_link_add(dev, &rdev->dev, DL_FLAG_STATELESS);
--
2.36.0