pm_runtime_get_if_in_use() does not only return nonzero values when the device is in use, it can return a negative errno too. And especially during resuming from system suspend, when runtime pm is not yet up again, -EAGAIN is being returned, so the subsequent pm_runtime_put() call results in a refcount underflow. Fix system-resume by handling -EAGAIN of pm_runtime_get_if_in_use(). Signed-off-by: Martin Kepplinger <martin.kepplinger@xxxxxxx> --- revision history ---------------- v2 (thank you Sakari and Laurent): * drop the other patch (the streaming-state in suspend/resume needs to be solved differently). * Sakari pointed out that many drivers are affected by this and that runtime-pm might need changes instead. I think this patch doesn't hurt and could serve as a reminder to do so. v1: initial submission: https://lore.kernel.org/linux-media/20230405092904.1129395-1-martin.kepplinger@xxxxxxx/ drivers/media/i2c/hi846.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/media/i2c/hi846.c b/drivers/media/i2c/hi846.c index 306dc35e925f..f8709cdf28b3 100644 --- a/drivers/media/i2c/hi846.c +++ b/drivers/media/i2c/hi846.c @@ -1353,7 +1353,8 @@ static int hi846_set_ctrl(struct v4l2_ctrl *ctrl) exposure_max); } - if (!pm_runtime_get_if_in_use(&client->dev)) + ret = pm_runtime_get_if_in_use(&client->dev); + if (!ret || ret == -EAGAIN) return 0; switch (ctrl->id) { -- 2.30.2
