| Message ID | 20240329101648.544155-2-u.kleine-koenig@pengutronix.de |
|---|---|
| State | Accepted |
| Headers | show |
| Series | pwm: Don't check pointer for being non-NULL after use | expand |
On Fri, Mar 29, 2024 at 5:17 AM Uwe Kleine-König <u.kleine-koenig@pengutronix.de> wrote: > > After assigning chip = pwm->chip; the compiler is free to assume that > pwm is non-NULL and so can optimize out the check for pwm against NULL. > > While it's probably a programming error to pass a NULL pointer to > pwm_put() this shouldn't be dropped without careful consideration and > wasn't intended. > > So assign chip only after the NULL check. > > Reported-by: David Lechner <dlechner@baylibre.com> > Fixes: d60bc2995732 ("pwm: Add a struct device to struct pwm_chip") > Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> > --- Reviewed-by: David Lechner <dlechner@baylibre.com>
Hello, On Fri, Mar 29, 2024 at 11:16:48AM +0100, Uwe Kleine-König wrote: > After assigning chip = pwm->chip; the compiler is free to assume that > pwm is non-NULL and so can optimize out the check for pwm against NULL. > > While it's probably a programming error to pass a NULL pointer to > pwm_put() this shouldn't be dropped without careful consideration and > wasn't intended. > > So assign chip only after the NULL check. > > Reported-by: David Lechner <dlechner@baylibre.com> > Fixes: d60bc2995732 ("pwm: Add a struct device to struct pwm_chip") > Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Applied to https://git.kernel.org/pub/scm/linux/kernel/git/ukleinek/linux.git pwm/for-next with David's Reviewed-by tag. Thanks Uwe
diff --git a/drivers/pwm/core.c b/drivers/pwm/core.c index 7f41ab087b98..54a62879fffa 100644 --- a/drivers/pwm/core.c +++ b/drivers/pwm/core.c @@ -1866,11 +1866,13 @@ EXPORT_SYMBOL_GPL(pwm_get); */ void pwm_put(struct pwm_device *pwm) { - struct pwm_chip *chip = pwm->chip; + struct pwm_chip *chip; if (!pwm) return; + chip = pwm->chip; + mutex_lock(&pwm_lock); /*
After assigning chip = pwm->chip; the compiler is free to assume that pwm is non-NULL and so can optimize out the check for pwm against NULL. While it's probably a programming error to pass a NULL pointer to pwm_put() this shouldn't be dropped without careful consideration and wasn't intended. So assign chip only after the NULL check. Reported-by: David Lechner <dlechner@baylibre.com> Fixes: d60bc2995732 ("pwm: Add a struct device to struct pwm_chip") Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> --- drivers/pwm/core.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) base-commit: d60bc2995732489766cbe650fc367e036ff6a0e0