cmdq_mbox_flush() calls pm_runtime_get_sync() and returns its error
code directly on failure. Per Documentation/power/runtime_pm.rst,
pm_runtime_get_sync() does not drop the usage counter on error, so the
caller is expected to release the reference itself. The early return
here skips that, leaking a runtime PM usage count on cmdq->mbox.dev on
every failed resume.
Fix it by calling pm_runtime_put_noidle() before returning, matching
the pattern used by other pm_runtime_get_sync() callers in the same
subsystem (e.g. drivers/i2c/busses/i2c-tegra.c, drivers/gnss/sirf.c).
Signed-off-by: manushprajwal <manushprajwal555@gmail.com>
---
drivers/mailbox/mtk-cmdq-mailbox.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/mailbox/mtk-cmdq-mailbox.c b/drivers/mailbox/mtk-cmdq-mailbox.c
index e523c84b4..2010aadbc 100644
--- a/drivers/mailbox/mtk-cmdq-mailbox.c
+++ b/drivers/mailbox/mtk-cmdq-mailbox.c
@@ -565,8 +565,10 @@ static int cmdq_mbox_flush(struct mbox_chan *chan, unsigned long timeout)
int ret;
ret = pm_runtime_get_sync(cmdq->mbox.dev);
- if (ret < 0)
+ if (ret < 0) {
+ pm_runtime_put_noidle(cmdq->mbox.dev);
return ret;
+ }
spin_lock_irqsave(&thread->chan->lock, flags);
if (list_empty(&thread->task_busy_list))
--
2.46.2.windows.1