linux-kernel - [PATCH 1/7] soundwire: bus: Do not forcibly disable child pm

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20220907101402.4685-2-rf@opensource.cirrus.com>
Date:   Wed, 7 Sep 2022 11:13:56 +0100
From:   Richard Fitzgerald <rf@...nsource.cirrus.com>
To:     <vkoul@...nel.org>, <yung-chuan.liao@...ux.intel.com>,
        <pierre-louis.bossart@...ux.intel.com>, <lgirdwood@...il.com>,
        <peter.ujfalusi@...ux.intel.com>,
        <ranjani.sridharan@...ux.intel.com>,
        <kai.vehmanen@...ux.intel.com>, <daniel.baluta@....com>,
        <sanyog.r.kale@...el.com>, <broonie@...nel.org>
CC:     <alsa-devel@...a-project.org>,
        <sound-open-firmware@...a-project.org>,
        <linux-kernel@...r.kernel.org>, <patches@...nsource.cirrus.com>,
        Richard Fitzgerald <rf@...nsource.cirrus.com>
Subject: [PATCH 1/7] soundwire: bus: Do not forcibly disable child pm_runtime

Do not call pm_runtime_disable() of a child driver in
sdw_delete_slave(). We really should never be trying to disable
another driver's pm_runtime - it is up to the child driver to
disable it or the core driver framework cleanup. The driver core
will runtime-resume a driver before calling its remove() so we
shouldn't break that.

The patch that introduced this is
commit dff70572e9a3 ("soundwire: bus: disable pm_runtime in sdw_slave_delete")
which says:

"prevent any race condition with the resume being executed after the
bus and slave devices are removed"

The actual problem is that the bus driver is shutting itself down before
the child drivers have been removed, which is the wrong way around (see
for example I2C and SPI drivers). If this is fixed, the bus driver will
still be operational when the driver framework runtime_resumes the child
drivers to remove them. Then the bus driver will remove() and can shut
down safely.

Also note that the child drivers are not necessarily idle when the bus
driver is removed, so disabling their pm_runtime and stopping the bus
might break more than only their remove().

Signed-off-by: Richard Fitzgerald <rf@...nsource.cirrus.com>
---
 drivers/soundwire/bus.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/soundwire/bus.c b/drivers/soundwire/bus.c
index 0bcc2d161eb9..99429892221b 100644
--- a/drivers/soundwire/bus.c
+++ b/drivers/soundwire/bus.c
@@ -151,8 +151,6 @@ static int sdw_delete_slave(struct device *dev, void *data)
 	struct sdw_slave *slave = dev_to_sdw_dev(dev);
 	struct sdw_bus *bus = slave->bus;

-	pm_runtime_disable(dev);
-
 	sdw_slave_debugfs_exit(slave);

 	mutex_lock(&bus->bus_lock);
-- 
2.30.2