]> git.ipfire.org Git - thirdparty/kernel/stable.git/commit
i2c: stm32f7: Do not prepare/unprepare clock during runtime suspend/resume
authorMarek Vasut <marex@denx.de>
Mon, 30 Sep 2024 19:27:41 +0000 (21:27 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 10 Oct 2024 10:00:46 +0000 (12:00 +0200)
commitfac3c9f7784e8184c0338e9f0877b81e55d3ef1c
tree140d14137b0534890b471ae554ea756c2cf211e6
parentbbefa2376a5f17a6199e7615ae285dff92d20b9c
i2c: stm32f7: Do not prepare/unprepare clock during runtime suspend/resume

commit 048bbbdbf85e5e00258dfb12f5e368f908801d7b upstream.

In case there is any sort of clock controller attached to this I2C bus
controller, for example Versaclock or even an AIC32x4 I2C codec, then
an I2C transfer triggered from the clock controller clk_ops .prepare
callback may trigger a deadlock on drivers/clk/clk.c prepare_lock mutex.

This is because the clock controller first grabs the prepare_lock mutex
and then performs the prepare operation, including its I2C access. The
I2C access resumes this I2C bus controller via .runtime_resume callback,
which calls clk_prepare_enable(), which attempts to grab the prepare_lock
mutex again and deadlocks.

Since the clock are already prepared since probe() and unprepared in
remove(), use simple clk_enable()/clk_disable() calls to enable and
disable the clock on runtime suspend and resume, to avoid hitting the
prepare_lock mutex.

Acked-by: Alain Volmat <alain.volmat@foss.st.com>
Signed-off-by: Marek Vasut <marex@denx.de>
Fixes: 4e7bca6fc07b ("i2c: i2c-stm32f7: add PM Runtime support")
Cc: <stable@vger.kernel.org> # v5.0+
Signed-off-by: Andi Shyti <andi.shyti@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/i2c/busses/i2c-stm32f7.c